WO2021241527A1 - Method for providing information for predicting effect of chemotherapy on non-small cell lung cancer and information provision kit, method for predicting effect of chemotherapy on non-small cell lung cancer, prediction system for predicting effect of chemotherapy on non-small cell lung cancer, and program and recording medium of prediction system - Google Patents
Method for providing information for predicting effect of chemotherapy on non-small cell lung cancer and information provision kit, method for predicting effect of chemotherapy on non-small cell lung cancer, prediction system for predicting effect of chemotherapy on non-small cell lung cancer, and program and recording medium of prediction system Download PDFInfo
- Publication number
- WO2021241527A1 WO2021241527A1 PCT/JP2021/019684 JP2021019684W WO2021241527A1 WO 2021241527 A1 WO2021241527 A1 WO 2021241527A1 JP 2021019684 W JP2021019684 W JP 2021019684W WO 2021241527 A1 WO2021241527 A1 WO 2021241527A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- chemotherapy
- abundance
- protein
- lung cancer
- small cell
- Prior art date
Links
Images
Classifications
-
- C—CHEMISTRY; METALLURGY
- C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
- C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
- C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
- C12Q1/02—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving viable microorganisms
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/48—Biological material, e.g. blood, urine; Haemocytometers
- G01N33/50—Chemical analysis of biological material, e.g. blood, urine; Testing involving biospecific ligand binding methods; Immunological testing
- G01N33/53—Immunoassay; Biospecific binding assay; Materials therefor
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/48—Biological material, e.g. blood, urine; Haemocytometers
- G01N33/50—Chemical analysis of biological material, e.g. blood, urine; Testing involving biospecific ligand binding methods; Immunological testing
- G01N33/53—Immunoassay; Biospecific binding assay; Materials therefor
- G01N33/574—Immunoassay; Biospecific binding assay; Materials therefor for cancer
Definitions
- the disclosure in this application is a method and information kit for providing information for predicting the effect of chemotherapy for non-small cell lung cancer, a method for predicting the effect of chemotherapy for non-small cell lung cancer, and a method for predicting the effect of chemotherapy for non-small cell lung cancer. It relates to a predictor, a program of the predictor, and a recording medium for predicting the effect of chemotherapy.
- Chemotherapy unlike molecular-targeted therapy, has no criteria for treatment selection such as gene mutation. Therefore, a method is known to measure the level of carcinogenic protein in the lysate of cancer cells isolated from a subject and predict the responsiveness of chemotherapy before chemotherapy (administration of anticancer drug). (See Patent Document 1).
- Patent Document 1 measures the level or activation state of the carcinogenic fusion protein contained in the sample collected from the subject. Therefore, there is a problem that the measurement procedure is complicated. It is desired to identify a biomarker (index) that can predict the effect of chemotherapy for lung cancer, especially non-small cell lung cancer, which has a high proportion, by a simpler method.
- the disclosure in this application is a method and information kit for providing information for predicting the effect of chemotherapy for non-small cell lung cancer, a method for predicting the effect of chemotherapy for non-small cell lung cancer, and non-small cell lung cancer.
- the present invention relates to a predictor, a program of the predictor, and a recording medium for predicting the effect of chemotherapy for small cell lung cancer.
- a method for providing information for predicting the effect of chemotherapy for non-small cell lung cancer is A step of measuring the abundance of at least Nucleolar protein 58 (Accession: Q9Y2X3) among the proteins in the biological sample of the subject. Including, how.
- the step of measuring the abundance measures the abundance of protein1 and protein2.
- the step of measuring the abundance is to measure the abundance of products1 to protein3.
- the step of measuring the abundance is to measure the abundance of products 1 to 11.
- the method according to (2) above. (6) The step of measuring the abundance measures the abundance of products1 to 36.
- Chemotherapy (A) Carboplatin and pemetrexed, (B) Pemetrexed and cisplatin, It is one of the combination therapies selected from The method according to any one of (1) to (6) above.
- (8) A method for predicting the effect of chemotherapy for non-small cell lung cancer. The method is In the biological sample of the subject, a prediction model constructed in advance based on the abundance of the protein according to any one of (1) to (7) above expressed in the biological sample of a non-small cell lung cancer patient was used.
- the abundance of protein expressed in the biological sample of the subject (hereinafter, may be referred to as "the amount of protein of the subject") is determined.
- Comprehensive protein abundance obtained by analyzing a sample by mass analysis or the like can be mentioned.
- only the abundance of the protein used as an index when constructing the prediction model may be used as the protein amount of the subject.
- subject includes those who do not have non-small cell lung cancer. In that case, when the protein of the subject is analyzed, it is assumed that the protein used as an index when constructing the prediction model does not exist. Therefore, in the present specification, when the description "abundance of protein expressed in the biological sample of the subject" is described, the case where the abundance is zero is also included.
- the prediction model is not particularly limited as long as the effect of chemotherapy can be predicted by comparing it with the amount of protein in the subject.
- it may be only the data showing the abundance of the protein as an index, or it may be a discriminant created by using statistical means.
- a threshold value may be further created as necessary, the amount of protein of the subject may be applied to the discriminant formula to calculate a score, and the effect of chemotherapy may be predicted by comparing with the threshold value.
- the embodiment of the information providing method is -A step of measuring the abundance of at least protein1 among the proteins in the biological sample of the subject, including.
- the step of measuring the abundance is not particularly limited as long as the abundance of protein1 can be measured, and may be a combination of at least one proteinin and protein1 selected from the above-mentioned protein2 to 36.
- the input unit 2 is not particularly limited as long as information regarding the amount of protein of the subject can be input to the prediction device 1, and examples thereof include a keyboard and USB. Further, the input unit 2 may use an internet line. For example, by transmitting and inputting information on the amount of subject protein measured at a remote hospital using an internet line to the prediction device 1 and sending the prediction result via the internet line, the subject at a remote hospital The effect of appropriate chemotherapy can be predicted.
- ⁇ patient ⁇ The predictive model was created for patients who met the registration criteria shown below. As shown in FIG. 1, blood was collected from 249 patients prior to the start of chemotherapy. (1) Cases diagnosed as non-flat epithelial non-small cell lung cancer by histology or cytology (2) Cases of clinical stages IIIA, IIIB, IV that are not subject to radical irradiation or surgical resection (3) Chemotherapy Cases without therapy (However, patients with postoperative adjuvant chemotherapy can be registered if there is an interval of 6 months or more after the last administration).
- Serum serum (sample) was separated by a routine method from blood collected before chemotherapy of patients for training data.
- the expressed proteins were comprehensively analyzed from the separated sera using a mass spectrometer (5600 manufactured by Siex Co., Ltd.). Information on the abundance of the analyzed protein and the effect of chemotherapy was associated.
- FIG. 6 is a graph showing an error rate when 1 to 483 proteins are selected.
- Error rate means (the number of examples of the number of evaluation samples that were incorrect) / (the total number of evaluation samples), and the lower the Error rate, the more preferable.
- selecting 1 to 483 proteins means that the proteins were selected in descending order of the number of selections. For example, selecting 1 to 36 proteins means a combination of the above M models, respectively.
- the correct answer rate was relatively high even when only protein 1 was used, but for example, there are at least a combination of protein 1 to 11 shown in the 11 models in which the Error rate is 10% or less (correct answer rate is 90% or more).
- the amount may be measured and the number of protein combinations may be increased as needed.
- the Error rate showed the smallest value (about 4.58%) in the case of the 36 proteins (proteins 1 to 36) shown in the 36 model, so in the following examples, the 36 proteins shown in the 36 model A predictive model was created using proteins.
- the created prediction model (discriminant) is shown below.
- S2Ns1x (score of protein1-borders1) + S2Ns2x (score of protein2-borders2) + ...
- S2Ns35x (score of protein35-borders35) + S2Ns36x (score of protein36-borders36)
- the risk score of each sample was calculated.
- the risk score can be calculated by performing the same calculation.
- the threshold value may be appropriately set based on the calculated risk score. For example, in Example 2 described later, the threshold value is set to 0, the score is ⁇ 0: good, and the score is ⁇ 0: better, but other values may be used.
- the correct answer rate (predicted to have a therapeutic effect) was 93.2% in the good group and 0% in the poor group, so the sensitivity was 93.2% and the specificity was 100%.
- the overall classification accuracy (overall specification accuracy) was 95.8%.
- FIG. 7 shows the ROC curve of the “good group patient” vs. the “poor group patient”, and the AUC (area under the curve: area under the concentration curve) was a very high value of 0.991.
- Example 2 [Contrast with verification group] As described above, since the prediction model prepared with the 36 proteins shown in Table 16 had high sensitivity and specificity, a comparison with the validation cohort was performed.
- the line graphs (Good and Poor) in FIG. 8 represent Kaplan-Meier curves. As is clear from FIG. 8, the overall survival time of the patients classified in the good group was longer than the overall survival time of the patients classified in the poor group. The median overall survival of 50% (0.5 on the vertical axis) was 25.7 months in the good group and 4.6 months in the poor group, which was about 5.6 times longer.
- the patient information shown in Table 19 below more specifically, age (75 years old or older vs. less than 75 years old), gender (male vs female), general condition of the patient "Performance Status: PS" (0/1 vs 2),. Cox regression analysis was performed considering information on smoking (none vs past / present), EGFR (other vs positive), stage (III vs IV), and stage (III vs recurrence). R was used as the analysis software. As shown in Table 19, the hazard ratio was 0.16, the 95% confidence interval was 0.09-0.30, and the p value was 9.54 ⁇ 10 -9 .
- the line graphs (Good and Poor) in FIG. 9 represent Kaplan-Meier curves. As is clear from FIG. 9, the progression-free survival of the patients classified in the good group was longer than the progression-free survival of the patients classified in the poor group. The median progression-free survival of 50% (0.5 on the vertical axis) was 6 months in the good group and 1.8 months in the poor group, which was about 3.3 times longer.
- the patient information shown in Table 20 below more specifically, age (75 years old or older vs. less than 75 years old), gender (male vs female), general condition of the patient "Performance Status: PS" (0/1 vs 2),. Cox regression analysis was performed considering information on smoking (none vs past / present), EDFR (other vs positive), stage (III vs IV), and stage (III vs recurrence). R was used as the analysis software. As shown in Table 20, the hazard ratio was 0.13, the 95% confidence interval was 0.08 to 0.24, and the p value was 1.67 ⁇ 10 -11 .
- the disclosure in this application can predict the effectiveness of chemotherapy for non-small cell lung cancer in advance. Therefore, it is useful for examination and research of lung cancer patients in research institutions such as medical institutions and university medical schools.
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Engineering & Computer Science (AREA)
- Immunology (AREA)
- Molecular Biology (AREA)
- Urology & Nephrology (AREA)
- Hematology (AREA)
- Biomedical Technology (AREA)
- Biotechnology (AREA)
- Biochemistry (AREA)
- Microbiology (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Organic Chemistry (AREA)
- Analytical Chemistry (AREA)
- Food Science & Technology (AREA)
- Proteomics, Peptides & Aminoacids (AREA)
- Cell Biology (AREA)
- Zoology (AREA)
- General Physics & Mathematics (AREA)
- Pathology (AREA)
- Wood Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Hospice & Palliative Care (AREA)
- Biophysics (AREA)
- Oncology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Genetics & Genomics (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The present invention addresses the problem of providing: a method for providing information for predicting the effect of chemotherapy on non-small cell lung cancer and an information provision kit; a method for predicting the effect of chemotherapy on non-small cell lung cancer; a prediction system for predicting the effect of chemotherapy on non-small cell lung cancer; and a program and a recording medium of the prediction system. This problem can be solved by a method for providing information for predicting the effect of chemotherapy on non-small cell lung cancer, said method comprising a step for measuring at least the amount of Nucleolar protein 58 (Accession: Q9Y2X3) among proteins in a biological sample derived from a subject.
Description
本出願における開示は、非小細胞肺がんの化学療法の効果を予測するための情報を提供する方法および情報提供用キット、非小細胞肺がんの化学療法の効果を予測する方法、非小細胞肺がんの化学療法の効果を予測する予測装置、予測装置のプログラムおよび記録媒体に関する。
The disclosure in this application is a method and information kit for providing information for predicting the effect of chemotherapy for non-small cell lung cancer, a method for predicting the effect of chemotherapy for non-small cell lung cancer, and a method for predicting the effect of chemotherapy for non-small cell lung cancer. It relates to a predictor, a program of the predictor, and a recording medium for predicting the effect of chemotherapy.
日本を含む殆どの先進諸国において、がんによる部位別死亡者数の中で、肺がんによる死亡は第1位を占めている。肺がんに対しては、様々な治療法の改良及び早期発見用の検査方法の改良が行われているが、日本においては毎年約75,000人の肺がん患者(以下、単に「患者」と記載することがある。)が死亡している。
In most developed countries including Japan, the number of deaths due to lung cancer is the highest among the number of deaths by site due to cancer. For lung cancer, various treatment methods have been improved and testing methods for early detection have been improved. In Japan, about 75,000 lung cancer patients (hereinafter referred to simply as "patients") are referred to every year. May be dead.)
患者に対する唯一治癒を望める治療法は手術治療であるが、初診時に手術を実施可能な割合は15%程度にとどまり、多くの患者が化学療法による治療を受けることとなる。近年の分子標的治療法の進歩により、遺伝子変異の有無を検査し、その結果に応じた治療薬が選択されることにより、中間生存期間の顕著な改善が得られるようになってきている。しかしながら、分子標的治療を実施したほとんどの患者は再増悪し、治療法の再選択を迫られることとなる。
Surgical treatment is the only treatment method that can be expected to cure patients, but the percentage of patients who can undergo surgery at the first visit is only about 15%, and many patients will be treated with chemotherapy. Recent advances in molecular-targeted therapies have led to significant improvements in intermediate survival by testing for the presence or absence of gene mutations and selecting therapeutic agents according to the results. However, most patients who receive molecular-targeted therapy will be exacerbated again and will be forced to reselect the treatment method.
再治療の際には、殺細胞性の治療薬を用いた化学療法が実施されることとなる。化学療法は、分子標的治療とは異なり、遺伝子変異などの治療法選択の基準がない。そのため、化学療法(抗がん剤の投与)前に、被検者から単離したがん細胞の溶解物中の発がん性タンパク質のレベルを測定し、化学療法の応答性を予測する方法が知られている(特許文献1参照)。
At the time of retreatment, chemotherapy using a cell-mediated therapeutic agent will be carried out. Chemotherapy, unlike molecular-targeted therapy, has no criteria for treatment selection such as gene mutation. Therefore, a method is known to measure the level of carcinogenic protein in the lysate of cancer cells isolated from a subject and predict the responsiveness of chemotherapy before chemotherapy (administration of anticancer drug). (See Patent Document 1).
しかしながら、特許文献1に記載の方法は、被検者から採取したサンプル中に含まれる発がん性融合タンパク質のレベルまたは活性化状態を測定している。したがって、測定手順が煩雑であるという問題がある。より簡便な方法により、被検者に対する肺がん、特に、割合が多い非小細胞肺がんの化学療法の効果を予測できるバイオマーカー(指標)の特定が望まれる。
However, the method described in Patent Document 1 measures the level or activation state of the carcinogenic fusion protein contained in the sample collected from the subject. Therefore, there is a problem that the measurement procedure is complicated. It is desired to identify a biomarker (index) that can predict the effect of chemotherapy for lung cancer, especially non-small cell lung cancer, which has a high proportion, by a simpler method.
本出願における開示は、上記従来の問題を解決するためになされた発明であり、鋭意研究を行ったところ、被検者の生体サンプル中のタンパク質の内、少なくともNucleolar protein 58の存在量が、非小細胞肺がんの化学療法の効果を予測するための情報として利用できることを新たに見出した。
The disclosure in this application is an invention made to solve the above-mentioned conventional problems, and as a result of diligent research, at least the abundance of Nucleolar protein 58 among the proteins in the biological sample of the subject is not. We have newly found that it can be used as information for predicting the effect of chemotherapy for small cell lung cancer.
すなわち、本出願における開示の目的は、非小細胞肺がんの化学療法の効果を予測するための情報を提供する方法および情報提供用キット、非小細胞肺がんの化学療法の効果を予測する方法、非小細胞肺がんの化学療法の効果を予測する予測装置、予測装置のプログラムおよび記録媒体を提供することである。
That is, the object of the disclosure in this application is a method for providing information for predicting the effect of chemotherapy for non-small cell lung cancer and a kit for providing information, a method for predicting the effect of chemotherapy for non-small cell lung cancer, and non-small cell lung cancer. It is to provide a predictor, a program of the predictor, and a recording medium for predicting the effect of chemotherapy for small cell lung cancer.
本出願における開示は、以下に示す、非小細胞肺がんの化学療法の効果を予測するための情報を提供する方法および情報提供用キット、非小細胞肺がんの化学療法の効果を予測する方法、非小細胞肺がんの化学療法の効果を予測する予測装置、予測装置のプログラムおよび記録媒体に関する。
The disclosure in this application is a method and information kit for providing information for predicting the effect of chemotherapy for non-small cell lung cancer, a method for predicting the effect of chemotherapy for non-small cell lung cancer, and non-small cell lung cancer. The present invention relates to a predictor, a program of the predictor, and a recording medium for predicting the effect of chemotherapy for small cell lung cancer.
(1)非小細胞肺がんの化学療法の効果を予測するための情報を提供する方法であって、
該方法が、
被検者の生体サンプル中のタンパク質の内、少なくともNucleolar protein 58(Accession:Q9Y2X3)の存在量を測定する工程、
を含む、方法。
(2)存在量を測定する工程が、請求項1に記載のタンパク質であるprotein1(Nucleolar protein 58)に加え、以下の表に記載のprotein2乃至36から選択される少なくとも1種以上のタンパク質の存在量を測定する、
上記(1)に記載の方法。
(3)存在量を測定する工程が、protein1およびprotein2の存在量を測定する、
上記(2)に記載の方法。
(4)存在量を測定する工程が、protein1乃至protein3の存在量を測定する、
上記(2)に記載の方法。
(5)存在量を測定する工程が、protein1乃至protein11の存在量を測定する、
上記(2)に記載の方法。
(6)存在量を測定する工程が、protein1乃至protein36の存在量を測定する、
上記(2)に記載の方法。
(7)化学療法が、
(a)カルボプラチンおよびペメトレキセド、
(b)ペメトレキセドおよびシスプラチン、
から選択される併用療法の1種である、
上記(1)~(6)の何れか一つに記載の方法。
(8)非小細胞肺がんの化学療法の効果を予測する方法であって、
該方法が、
非小細胞肺がん患者の生体サンプル中で発現している上記(1)~(7)の何れか一つに記載のタンパク質の存在量に基づき予め構築した予測モデルに、被検者の生体サンプル中で発現しているタンパク質の存在量を当てはめる工程と、
予測モデルに当てはめた被検者のタンパク質の存在量から、化学療法の効果を予測する予測工程と、
を含む、方法。
(9)予め構築した予測モデルが、
非小細胞肺がん患者の生体サンプル中で発現している上記(1)~(7)の何れか一つに記載のタンパク質の存在量を、統計的手段を用いて作成した判別式および閾値であり、
予測工程が、
被検者の生体サンプル中で発現しているタンパク質の存在量を判別式に当てはめスコアを算出し閾値と比較することで、
化学療法の効果あり、または、化学療法の効果無し、を予測する、
上記(8)に記載の方法。
(10)非小細胞肺がん患者の生体サンプル中で発現している上記(1)~(7)の何れか一つに記載のタンパク質の存在量に基づき予め構築した予測モデルを少なくとも格納した記憶部と、
被検者の生体サンプル中で発現しているタンパク質の存在量を記憶部に記憶された予測モデルに当てはめることで、被検者の化学療法の効果を予測する演算部と、
を含む、非小細胞肺がんの化学療法の効果を予測する予測装置。
(11)コンピュータを、上記(10)に記載の予測装置として機能させるためのプログラム。
(12)上記(11)に記載のプログラムを記録したコンピュータ読み取り可能な記録媒体。
(13)被検者の生体サンプル中のタンパク質の内、少なくともNucleolar protein 58(Accession:Q9Y2X3)を特異的に認識する抗体を含む、
非小細胞肺がんの化学療法の効果を予測するための情報提供用キット。 (1) A method for providing information for predicting the effect of chemotherapy for non-small cell lung cancer.
The method is
A step of measuring the abundance of at least Nucleolar protein 58 (Accession: Q9Y2X3) among the proteins in the biological sample of the subject.
Including, how.
(2) In the step of measuring the abundance, in addition to the protein 1 (Nucleolar protein 58) according toclaim 1, the presence of at least one protein selected from the proteins 2 to 36 described in the following table. Measure the amount,
The method according to (1) above.
(3) The step of measuring the abundance measures the abundance of protein1 and protein2.
The method according to (2) above.
(4) The step of measuring the abundance is to measure the abundance of products1 to protein3.
The method according to (2) above.
(5) The step of measuring the abundance is to measure the abundance ofproducts 1 to 11.
The method according to (2) above.
(6) The step of measuring the abundance measures the abundance of products1 to 36.
The method according to (2) above.
(7) Chemotherapy
(A) Carboplatin and pemetrexed,
(B) Pemetrexed and cisplatin,
It is one of the combination therapies selected from
The method according to any one of (1) to (6) above.
(8) A method for predicting the effect of chemotherapy for non-small cell lung cancer.
The method is
In the biological sample of the subject, a prediction model constructed in advance based on the abundance of the protein according to any one of (1) to (7) above expressed in the biological sample of a non-small cell lung cancer patient was used. The process of applying the abundance of the protein expressed in
A prediction process that predicts the effect of chemotherapy from the abundance of protein in the subject applied to the prediction model,
Including, how.
(9) The forecast model built in advance is
It is a discriminant and a threshold value prepared by using statistical means for the abundance of the protein according to any one of (1) to (7) above expressed in a biological sample of a non-small cell lung cancer patient. ,
The prediction process is
By applying the abundance of protein expressed in the biological sample of the subject to the discriminant formula, calculating the score, and comparing it with the threshold value,
Predict whether chemotherapy is effective or not.
The method according to (8) above.
(10) A storage unit containing at least a predictive model constructed in advance based on the abundance of the protein according to any one of (1) to (7) above expressed in a biological sample of a non-small cell lung cancer patient. When,
By applying the abundance of protein expressed in the subject's biological sample to the prediction model stored in the memory, the calculation unit that predicts the effect of the subject's chemotherapy, and the calculation unit.
Predictors that predict the effects of chemotherapy for non-small cell lung cancer, including.
(11) A program for making a computer function as the prediction device according to the above (10).
(12) A computer-readable recording medium on which the program described in (11) above is recorded.
(13) Among the proteins in the biological sample of the subject, an antibody that specifically recognizes at least Nucleolar protein 58 (Accession: Q9Y2X3) is contained.
An informational kit for predicting the effects of chemotherapy for non-small cell lung cancer.
該方法が、
被検者の生体サンプル中のタンパク質の内、少なくともNucleolar protein 58(Accession:Q9Y2X3)の存在量を測定する工程、
を含む、方法。
(2)存在量を測定する工程が、請求項1に記載のタンパク質であるprotein1(Nucleolar protein 58)に加え、以下の表に記載のprotein2乃至36から選択される少なくとも1種以上のタンパク質の存在量を測定する、
上記(1)に記載の方法。
上記(2)に記載の方法。
(4)存在量を測定する工程が、protein1乃至protein3の存在量を測定する、
上記(2)に記載の方法。
(5)存在量を測定する工程が、protein1乃至protein11の存在量を測定する、
上記(2)に記載の方法。
(6)存在量を測定する工程が、protein1乃至protein36の存在量を測定する、
上記(2)に記載の方法。
(7)化学療法が、
(a)カルボプラチンおよびペメトレキセド、
(b)ペメトレキセドおよびシスプラチン、
から選択される併用療法の1種である、
上記(1)~(6)の何れか一つに記載の方法。
(8)非小細胞肺がんの化学療法の効果を予測する方法であって、
該方法が、
非小細胞肺がん患者の生体サンプル中で発現している上記(1)~(7)の何れか一つに記載のタンパク質の存在量に基づき予め構築した予測モデルに、被検者の生体サンプル中で発現しているタンパク質の存在量を当てはめる工程と、
予測モデルに当てはめた被検者のタンパク質の存在量から、化学療法の効果を予測する予測工程と、
を含む、方法。
(9)予め構築した予測モデルが、
非小細胞肺がん患者の生体サンプル中で発現している上記(1)~(7)の何れか一つに記載のタンパク質の存在量を、統計的手段を用いて作成した判別式および閾値であり、
予測工程が、
被検者の生体サンプル中で発現しているタンパク質の存在量を判別式に当てはめスコアを算出し閾値と比較することで、
化学療法の効果あり、または、化学療法の効果無し、を予測する、
上記(8)に記載の方法。
(10)非小細胞肺がん患者の生体サンプル中で発現している上記(1)~(7)の何れか一つに記載のタンパク質の存在量に基づき予め構築した予測モデルを少なくとも格納した記憶部と、
被検者の生体サンプル中で発現しているタンパク質の存在量を記憶部に記憶された予測モデルに当てはめることで、被検者の化学療法の効果を予測する演算部と、
を含む、非小細胞肺がんの化学療法の効果を予測する予測装置。
(11)コンピュータを、上記(10)に記載の予測装置として機能させるためのプログラム。
(12)上記(11)に記載のプログラムを記録したコンピュータ読み取り可能な記録媒体。
(13)被検者の生体サンプル中のタンパク質の内、少なくともNucleolar protein 58(Accession:Q9Y2X3)を特異的に認識する抗体を含む、
非小細胞肺がんの化学療法の効果を予測するための情報提供用キット。 (1) A method for providing information for predicting the effect of chemotherapy for non-small cell lung cancer.
The method is
A step of measuring the abundance of at least Nucleolar protein 58 (Accession: Q9Y2X3) among the proteins in the biological sample of the subject.
Including, how.
(2) In the step of measuring the abundance, in addition to the protein 1 (Nucleolar protein 58) according to
The method according to (1) above.
The method according to (2) above.
(4) The step of measuring the abundance is to measure the abundance of products1 to protein3.
The method according to (2) above.
(5) The step of measuring the abundance is to measure the abundance of
The method according to (2) above.
(6) The step of measuring the abundance measures the abundance of products1 to 36.
The method according to (2) above.
(7) Chemotherapy
(A) Carboplatin and pemetrexed,
(B) Pemetrexed and cisplatin,
It is one of the combination therapies selected from
The method according to any one of (1) to (6) above.
(8) A method for predicting the effect of chemotherapy for non-small cell lung cancer.
The method is
In the biological sample of the subject, a prediction model constructed in advance based on the abundance of the protein according to any one of (1) to (7) above expressed in the biological sample of a non-small cell lung cancer patient was used. The process of applying the abundance of the protein expressed in
A prediction process that predicts the effect of chemotherapy from the abundance of protein in the subject applied to the prediction model,
Including, how.
(9) The forecast model built in advance is
It is a discriminant and a threshold value prepared by using statistical means for the abundance of the protein according to any one of (1) to (7) above expressed in a biological sample of a non-small cell lung cancer patient. ,
The prediction process is
By applying the abundance of protein expressed in the biological sample of the subject to the discriminant formula, calculating the score, and comparing it with the threshold value,
Predict whether chemotherapy is effective or not.
The method according to (8) above.
(10) A storage unit containing at least a predictive model constructed in advance based on the abundance of the protein according to any one of (1) to (7) above expressed in a biological sample of a non-small cell lung cancer patient. When,
By applying the abundance of protein expressed in the subject's biological sample to the prediction model stored in the memory, the calculation unit that predicts the effect of the subject's chemotherapy, and the calculation unit.
Predictors that predict the effects of chemotherapy for non-small cell lung cancer, including.
(11) A program for making a computer function as the prediction device according to the above (10).
(12) A computer-readable recording medium on which the program described in (11) above is recorded.
(13) Among the proteins in the biological sample of the subject, an antibody that specifically recognizes at least Nucleolar protein 58 (Accession: Q9Y2X3) is contained.
An informational kit for predicting the effects of chemotherapy for non-small cell lung cancer.
なお、上記のタンパク質のAccessionは、全てUniProt(https://www.uniprot.org)の番号である。以下、Accessionは省略することがある。
Note that the Accessions for the above proteins are all UniProt (https://www.uniprot.org) numbers. Hereinafter, Accession may be omitted.
被検者の生体サンプル中のタンパク質の内、少なくともNucleolar protein 58の存在量を情報とすることで、非小細胞肺がんの化学療法の効果を予測できる。
The effect of chemotherapy for non-small cell lung cancer can be predicted by using information on the abundance of at least Nucleolar protein 58 among the proteins in the biological sample of the subject.
以下に、非小細胞肺がんの化学療法の効果を予測するための情報を提供する方法(以下、「情報提供方法」と記載することがある。)および情報提供用キット、非小細胞肺がんの化学療法の効果を予測する方法(以下、「予測方法」と記載することがある。)、非小細胞肺がんの化学療法の効果を予測する予測装置(以下、「予測装置」と記載することがある。)、予測装置のプログラム(以下、「プログラム」と記載することがある)および記録媒体について詳しく説明する。
Below, a method for providing information for predicting the effect of chemotherapy for non-small cell lung cancer (hereinafter, may be referred to as "information providing method"), an information providing kit, and chemistry for non-small cell lung cancer. A method for predicting the effect of therapy (hereinafter, may be referred to as "prediction method"), and a prediction device for predicting the effect of chemotherapy for non-small cell lung cancer (hereinafter, may be referred to as "prediction device"). ), The program of the predictor (hereinafter, may be referred to as "program") and the recording medium will be described in detail.
(予測方法の実施形態)
図1および図2を参照し、予測方法の実施形態について説明する。図1は、予測方法の実施形態に用いる「予測モデル」の作成手順、図2は作成した予測モデルの検証手順の概略を示す図である。なお、図1および図2中の数値(N;サンプル数)は、後述する実施例で用いた具体的な人数である。予測モデルの作成に必要な患者のサンプル数が、図1および図2に記載の数値と異なってもよいことは言うまでもない。 (Embodiment of Prediction Method)
An embodiment of the prediction method will be described with reference to FIGS. 1 and 2. FIG. 1 is a diagram showing an outline of a procedure for creating a “prediction model” used in an embodiment of a prediction method, and FIG. 2 is a diagram showing an outline of a procedure for verifying the created prediction model. The numerical values (N; number of samples) in FIGS. 1 and 2 are specific numbers used in the examples described later. It goes without saying that the number of patient samples required to create a predictive model may differ from the numbers shown in FIGS. 1 and 2.
図1および図2を参照し、予測方法の実施形態について説明する。図1は、予測方法の実施形態に用いる「予測モデル」の作成手順、図2は作成した予測モデルの検証手順の概略を示す図である。なお、図1および図2中の数値(N;サンプル数)は、後述する実施例で用いた具体的な人数である。予測モデルの作成に必要な患者のサンプル数が、図1および図2に記載の数値と異なってもよいことは言うまでもない。 (Embodiment of Prediction Method)
An embodiment of the prediction method will be described with reference to FIGS. 1 and 2. FIG. 1 is a diagram showing an outline of a procedure for creating a “prediction model” used in an embodiment of a prediction method, and FIG. 2 is a diagram showing an outline of a procedure for verifying the created prediction model. The numerical values (N; number of samples) in FIGS. 1 and 2 are specific numbers used in the examples described later. It goes without saying that the number of patient samples required to create a predictive model may differ from the numbers shown in FIGS. 1 and 2.
先ず、予測モデルの作成手順について説明する。予測モデルの作成には、非小細胞肺がん患者(以下、単に「患者」と記載することがある。)の生体サンプルを集める。図1では、生体サンプルとして血清を用いた例が記載されているが、生体サンプルは、非小細胞肺がん由来のタンパク質が分泌されていれば特に制限はない。例えば、全血、血清、血漿、尿、痰、気管支洗浄液、涙液、乳頭吸引液、リンパ液、唾液、微細針吸引液(FNA)、単離したがん細胞及びそれらの組み合わせが挙げられる。なお、本明細書において、「非小細胞肺がん」とは、小細胞肺がん以外の肺がんを意味する。例えば、扁平上皮がん、大細胞がん、腺がんが挙げられる。
First, the procedure for creating a prediction model will be explained. To create a predictive model, a biological sample of a non-small cell lung cancer patient (hereinafter, may be simply referred to as "patient") is collected. In FIG. 1, an example in which serum is used as a biological sample is described, but the biological sample is not particularly limited as long as a protein derived from non-small cell lung cancer is secreted. Examples include whole blood, serum, plasma, urine, sputum, bronchial lavage fluid, tear fluid, papillary aspirate, lymph, saliva, fine needle aspirate (FNA), isolated cancer cells and combinations thereof. In addition, in this specification, "non-small cell lung cancer" means lung cancer other than small cell lung cancer. Examples include squamous cell carcinoma, large cell carcinoma, and adenocarcinoma.
図1に示す“Classifier construction”では、患者の採取血液から血清(サンプル)を分離し、質量分析により、血清中のタンパク質の存在量を測定する(N=249)。次に、予測モデルの作成に好ましくないと判断した患者を除き、教師群(training cohort)を決定する。次に、教師群に対して化学療法(CbP)を実施する。図1では、化学療法として、カルボプラチン(carboplatin:CBDCA)およびペメトレキセド(pemetrexed:PEM)を併用した例が記載されている。
In the "Classifier measurement" shown in FIG. 1, serum (sample) is separated from the collected blood of a patient, and the abundance of protein in the serum is measured by mass spectrometry (N = 249). Next, the teacher group (training cohort) is determined except for the patients who are judged to be unfavorable for creating a predictive model. Next, chemotherapy (CbP) is given to the teachers. FIG. 1 describes an example of combined use of carboplatin (CBDCA) and pemetrexed (PEM) as chemotherapy.
なお、化学療法に用いる化合物は、カルボプラチンおよびペメトレキセドの併用に限定されない。例えば、カルボプラチンと同様の白金錯体であるシスプラチン(cisplatin:CDDP)をペメトレキセドと併用してもよい。シスプラチンは、以下の式(3)で表される化合物である。
The compound used for chemotherapy is not limited to the combined use of carboplatin and pemetrexed. For example, cisplatin (CDDP), which is a platinum complex similar to carboplatin, may be used in combination with pemetrexed. Cisplatin is a compound represented by the following formula (3).
次に、化学療法を実施した後に、患者のCT画像を撮影する。そして、CT画像に基づき、主治医から独立した内科医および放射線科医によって、RECISTガイドライン version 1.1に基づき化学療法の効果の有無を評価する。治療効果あり、または、治療効果無し、と評価された患者(N=96)の化学療法前の血清を質量分析と関連付ける。次に、治療効果あり、および、治療効果無し、と評価された患者の血清中のタンパク質の存在量の増減が大きなタンパク質を特定し、予測モデルの作成に用いる分類子を構築する(Classifier construction)。
Next, after performing chemotherapy, take a CT image of the patient. Then, based on the CT image, the presence or absence of the effect of chemotherapy is evaluated by a physician and a radiologist independent of the attending physician based on the RECIST guideline version 1.1. Pre-chemotherapy sera of patients (N = 96) evaluated as therapeutic or ineffective are associated with mass spectrometry. Next, identify proteins with a large increase or decrease in the serum abundance of proteins evaluated as having a therapeutic effect and no therapeutic effect, and construct a classifier used for creating a predictive model (Classifier constraint). ..
図2に示す“Classifier validation”では、図1に示す“Classifier construction”において予測モデル用として特定したタンパク質の優位性を検証する。より具体的には、教師群とは別の患者(検証群)から、化学療法を実施する前の血清を採取し、質量分析することで血清中のタンパク質の存在量を測定する。そして、“Classifier construction”で作成した予測モデルに、検証群の患者血清中のタンパク質の存在量を当てはめ、検証群の患者の化学療法の有効性を判定する。判定結果と患者(Validation cohort;N=94)の化学療法実施後の実際の治療効果とを対比することで、作成した予測モデルの優位性の検証を行う。
In the "Classifier validation" shown in FIG. 2, the superiority of the protein specified for the predictive model in the "Classifier context" shown in FIG. 1 is verified. More specifically, serum before chemotherapy is collected from a patient (verification group) different from the teacher group, and mass spectrometry is performed to measure the abundance of protein in the serum. Then, the abundance of protein in the serum of the patients in the verification group is applied to the prediction model created by the "Classifier constraint" to determine the effectiveness of chemotherapy for the patients in the verification group. By comparing the judgment result with the actual therapeutic effect of the patient (Validation cohort; N = 94) after chemotherapy, the superiority of the created predictive model will be verified.
後述する実施例に示すように、予測モデルは、以下の表に示す36種類のタンパク質(protein1~36)の内、少なくともprotein1の存在量を指標として作成すればよい。また、指標とするタンパク質の種類が多いほど、予測精度が高くなる。したがって、protein1に加え、protein2乃至36から選択される少なくとも1種以上のタンパク質の存在量の組み合わせを指標としてもよい。
As shown in the examples described later, the prediction model may be created using at least the abundance of protein1 among the 36 types of proteins (proteinin1 to 36) shown in the table below as an index. In addition, the more types of protein used as an index, the higher the prediction accuracy. Therefore, in addition to protein1, a combination of abundances of at least one protein selected from protein2 to 36 may be used as an index.
なお、後記する実施例に示すとおり、上記表のprotein1乃至36は、番号が小さいほど選択回数が多いことから、指標として信頼性が高いと考えられる。したがって、protein1乃至36から1種以上を選択してprotein1と組み合して指標とする際には、例えば、以下の組み合わせが挙げられる。
As shown in the examples described later, since the smaller the number, the larger the number of selections of products 1 to 36 in the above table, it is considered that the reliability is high as an index. Therefore, when one or more kinds are selected from protein 1 to 36 and combined with protein 1 to be used as an index, for example, the following combinations can be mentioned.
・protein1およびprotein2(2種類の組み合わせ)
・protein1乃至protein3(3種類の組み合わせ)
・protein1乃至protein4(4種類の組み合わせ)
・protein1乃至protein5(5種類の組み合わせ)
・protein1乃至protein6(6種類の組み合わせ)
・protein1乃至protein7(7種類の組み合わせ)
・protein1乃至protein8(8種類の組み合わせ)
・protein1乃至protein9(9種類の組み合わせ)
・protein1乃至protein10(10種類の組み合わせ)
・protein1乃至protein11(11種類の組み合わせ)
・protein1乃至protein12(12種類の組み合わせ)
・protein1乃至protein12(13種類の組み合わせ)
・protein1乃至protein14(14種類の組み合わせ)
・protein1乃至protein15(15種類の組み合わせ)
・protein1乃至protein16(16種類の組み合わせ)
・protein1乃至protein17(17種類の組み合わせ)
・protein1乃至protein18(18種類の組み合わせ)
・protein1乃至protein19(19種類の組み合わせ)
・protein1乃至protein20(20種類の組み合わせ)
・protein1乃至protein21(21種類の組み合わせ)
・protein1乃至protein22(22種類の組み合わせ)
・protein1乃至protein23(23種類の組み合わせ)
・protein1乃至protein24(24種類の組み合わせ)
・protein1乃至protein25(25種類の組み合わせ)
・protein1乃至protein26(26種類の組み合わせ)
・protein1乃至protein27(27種類の組み合わせ)
・protein1乃至protein28(28種類の組み合わせ)
・protein1乃至protein29(29種類の組み合わせ)
・protein1乃至protein30(30種類の組み合わせ)
・protein1乃至protein31(31種類の組み合わせ)
・protein1乃至protein32(32種類の組み合わせ)
・protein1乃至protein33(33種類の組み合わせ)
・protein1乃至protein34(34種類の組み合わせ)
・protein1乃至protein35(35種類の組み合わせ)
・protein1乃至protein36(36種類の組み合わせ) ・ Protein1 and protein2 (combination of two types)
・ Protein1 to protein3 (combination of 3 types)
・ Protein1 to protein4 (4 types of combinations)
・ Protein1 to protein5 (combination of 5 types)
・ Protein1 to protein6 (6 types of combinations)
・ Protein1 to protein7 (7 types of combinations)
・ Protein1 to protein8 (8 types of combinations)
・ Protein1 to protein9 (9 types of combinations)
・ Protein1 to protein10 (10 types of combinations)
・ Protein1 to protein11 (11 types of combinations)
・ Protein1 to protein12 (12 types of combinations)
・ Protein1 to protein12 (13 types of combinations)
・ Protein1 to protein14 (14 types of combinations)
・ Protein1 to protein15 (15 types of combinations)
・ Protein1 to protein16 (16 types of combinations)
・ Protein1 to protein17 (17 types of combinations)
・ Protein1 to protein18 (18 combinations)
・ Protein1 to protein19 (19 types of combinations)
・ Protein1 to protein20 (20 types of combinations)
・ Protein1 to protein21 (21 types of combinations)
・ Protein1 to protein22 (22 types of combinations)
・ Protein1 to protein23 (23 types of combinations)
・ Protein1 to protein24 (24 types of combinations)
・ Protein1 to protein25 (25 types of combinations)
・ Protein1 to protein26 (26 types of combinations)
・ Protein1 to protein27 (27 types of combinations)
・ Protein1 to protein28 (28 types of combinations)
・ Protein1 to protein29 (29 types of combinations)
・ Protein1 to protein30 (30 types of combinations)
・ Protein1 to protein31 (31 types of combinations)
-Protein1 to protein32 (32 types of combinations)
・ Protein1 to protein33 (33 types of combinations)
・ Protein1 to protein34 (34 types of combinations)
・ Protein1 to protein35 (35 types of combinations)
・ Protein1 to protein36 (36 types of combinations)
・protein1乃至protein3(3種類の組み合わせ)
・protein1乃至protein4(4種類の組み合わせ)
・protein1乃至protein5(5種類の組み合わせ)
・protein1乃至protein6(6種類の組み合わせ)
・protein1乃至protein7(7種類の組み合わせ)
・protein1乃至protein8(8種類の組み合わせ)
・protein1乃至protein9(9種類の組み合わせ)
・protein1乃至protein10(10種類の組み合わせ)
・protein1乃至protein11(11種類の組み合わせ)
・protein1乃至protein12(12種類の組み合わせ)
・protein1乃至protein12(13種類の組み合わせ)
・protein1乃至protein14(14種類の組み合わせ)
・protein1乃至protein15(15種類の組み合わせ)
・protein1乃至protein16(16種類の組み合わせ)
・protein1乃至protein17(17種類の組み合わせ)
・protein1乃至protein18(18種類の組み合わせ)
・protein1乃至protein19(19種類の組み合わせ)
・protein1乃至protein20(20種類の組み合わせ)
・protein1乃至protein21(21種類の組み合わせ)
・protein1乃至protein22(22種類の組み合わせ)
・protein1乃至protein23(23種類の組み合わせ)
・protein1乃至protein24(24種類の組み合わせ)
・protein1乃至protein25(25種類の組み合わせ)
・protein1乃至protein26(26種類の組み合わせ)
・protein1乃至protein27(27種類の組み合わせ)
・protein1乃至protein28(28種類の組み合わせ)
・protein1乃至protein29(29種類の組み合わせ)
・protein1乃至protein30(30種類の組み合わせ)
・protein1乃至protein31(31種類の組み合わせ)
・protein1乃至protein32(32種類の組み合わせ)
・protein1乃至protein33(33種類の組み合わせ)
・protein1乃至protein34(34種類の組み合わせ)
・protein1乃至protein35(35種類の組み合わせ)
・protein1乃至protein36(36種類の組み合わせ) ・ Protein1 and protein2 (combination of two types)
・ Protein1 to protein3 (combination of 3 types)
・ Protein1 to protein4 (4 types of combinations)
・ Protein1 to protein5 (combination of 5 types)
・ Protein1 to protein6 (6 types of combinations)
・ Protein1 to protein7 (7 types of combinations)
・ Protein1 to protein8 (8 types of combinations)
・ Protein1 to protein9 (9 types of combinations)
・ Protein1 to protein10 (10 types of combinations)
・ Protein1 to protein11 (11 types of combinations)
・ Protein1 to protein12 (12 types of combinations)
・ Protein1 to protein12 (13 types of combinations)
・ Protein1 to protein14 (14 types of combinations)
・ Protein1 to protein15 (15 types of combinations)
・ Protein1 to protein16 (16 types of combinations)
・ Protein1 to protein17 (17 types of combinations)
・ Protein1 to protein18 (18 combinations)
・ Protein1 to protein19 (19 types of combinations)
・ Protein1 to protein20 (20 types of combinations)
・ Protein1 to protein21 (21 types of combinations)
・ Protein1 to protein22 (22 types of combinations)
・ Protein1 to protein23 (23 types of combinations)
・ Protein1 to protein24 (24 types of combinations)
・ Protein1 to protein25 (25 types of combinations)
・ Protein1 to protein26 (26 types of combinations)
・ Protein1 to protein27 (27 types of combinations)
・ Protein1 to protein28 (28 types of combinations)
・ Protein1 to protein29 (29 types of combinations)
・ Protein1 to protein30 (30 types of combinations)
・ Protein1 to protein31 (31 types of combinations)
-Protein1 to protein32 (32 types of combinations)
・ Protein1 to protein33 (33 types of combinations)
・ Protein1 to protein34 (34 types of combinations)
・ Protein1 to protein35 (35 types of combinations)
・ Protein1 to protein36 (36 types of combinations)
なお、上記の組み合わせは単なる例示である。例えば、N種類の組み合わせと記載した場合には、protein1に加え、protein2乃至36から選択した任意のN-1種類のタンパク質の組み合わせであってもよい。
The above combination is just an example. For example, when described as a combination of N types, it may be a combination of any N-1 type of protein selected from protein2 to 36 in addition to protein1.
タンパク質の存在量を測定する際には、必要に応じて、ノーマライズにより補正してもよい。ノーマライズは、患者サンプル毎に検出されたタンパク質全体の測定値から中央値を計算し、その中央値で個々の測定値を割ることで実施できる。
When measuring the abundance of protein, it may be corrected by normalization if necessary. Normalization can be performed by calculating the median from the overall protein measurements detected for each patient sample and dividing the individual measurements by the median.
予測方法の実施形態は、
(1)非小細胞肺がん患者の生体サンプル中で発現している指標となるタンパク質の存在量に基づき予め構築した予測モデルに、被検者の生体サンプル中で発現しているタンパク質の存在量を当てはめる工程と、
(2)予測モデルに当てはめた被検者のタンパク質の存在量から、化学療法の効果を予測する予測工程と、
を含んでいる。 The embodiment of the prediction method is
(1) The abundance of the protein expressed in the biological sample of the subject is added to the prediction model constructed in advance based on the abundance of the protein as an index expressed in the biological sample of the non-small cell lung cancer patient. The process of applying and
(2) A prediction process for predicting the effect of chemotherapy from the abundance of protein in the subject applied to the prediction model.
Includes.
(1)非小細胞肺がん患者の生体サンプル中で発現している指標となるタンパク質の存在量に基づき予め構築した予測モデルに、被検者の生体サンプル中で発現しているタンパク質の存在量を当てはめる工程と、
(2)予測モデルに当てはめた被検者のタンパク質の存在量から、化学療法の効果を予測する予測工程と、
を含んでいる。 The embodiment of the prediction method is
(1) The abundance of the protein expressed in the biological sample of the subject is added to the prediction model constructed in advance based on the abundance of the protein as an index expressed in the biological sample of the non-small cell lung cancer patient. The process of applying and
(2) A prediction process for predicting the effect of chemotherapy from the abundance of protein in the subject applied to the prediction model.
Includes.
上記(1)に記載の工程において、予測モデルを構築する際の指標となるタンパク質は、上記に例示した、
・protein1、或いは、
・protein2乃至36から選択される少なくとも1種のproteinおよびprotein1の組み合わせ、
である。 In the step described in (1) above, the proteins that serve as indicators for constructing a predictive model are exemplified above.
・ Protein1 or
A combination of at least one protein andprotein 1 selected from proteins 2 to 36,
Is.
・protein1、或いは、
・protein2乃至36から選択される少なくとも1種のproteinおよびprotein1の組み合わせ、
である。 In the step described in (1) above, the proteins that serve as indicators for constructing a predictive model are exemplified above.
・ Protein1 or
A combination of at least one protein and
Is.
上記(1)および(2)に記載の工程において、被検者の生体サンプル中で発現しているタンパク質の存在量(以下、「被検者タンパク質量」と記載することがある。)は、サンプルを質量分析等により解析した網羅的なタンパク質の存在量が挙げられる。或いは、抗体等を用いて、予測モデルを構築する際に指標として用いたタンパク質の存在量のみを被検者タンパク質量としてもよい。なお、「被検者」と記載した場合には、非小細胞肺がんに罹患していない者も含まれる。その場合、被検者のタンパク質を解析すると、予測モデルを構築する際に指標としたタンパク質が存在しない場合も想定される。したがって、本明細書において、「被検者の生体サンプル中で発現しているタンパク質の存在量」と記載した場合、存在量がゼロの場合も含まれる。
In the steps described in (1) and (2) above, the abundance of protein expressed in the biological sample of the subject (hereinafter, may be referred to as "the amount of protein of the subject") is determined. Comprehensive protein abundance obtained by analyzing a sample by mass analysis or the like can be mentioned. Alternatively, using an antibody or the like, only the abundance of the protein used as an index when constructing the prediction model may be used as the protein amount of the subject. The term "subject" includes those who do not have non-small cell lung cancer. In that case, when the protein of the subject is analyzed, it is assumed that the protein used as an index when constructing the prediction model does not exist. Therefore, in the present specification, when the description "abundance of protein expressed in the biological sample of the subject" is described, the case where the abundance is zero is also included.
予測モデルは、被検者タンパク質量と対比することで、化学療法の効果を予測することができれば特に制限はない。例えば、指標となるタンパク質の存在量を表すデータのみであってもよいし、統計的手段を用いて作成した判別式であってもよい。判別式の場合、更に必要に応じて閾値を作成し、被検者タンパク質量を判別式に当てはめスコアを算出し、閾値と比較することで化学療法の効果を予測してもよい。
The prediction model is not particularly limited as long as the effect of chemotherapy can be predicted by comparing it with the amount of protein in the subject. For example, it may be only the data showing the abundance of the protein as an index, or it may be a discriminant created by using statistical means. In the case of a discriminant, a threshold value may be further created as necessary, the amount of protein of the subject may be applied to the discriminant formula to calculate a score, and the effect of chemotherapy may be predicted by comparing with the threshold value.
(情報提供方法の実施形態)
次に、情報提供方法の実施形態について説明する。情報提供方法の実施形態は、
・被検者の生体サンプル中のタンパク質の内、少なくともprotein1の存在量を測定する工程、
を含む。存在量を測定する工程は、少なくともprotein1の存在量を測定できれば特に制限はなく、上記protein2乃至36から選択される少なくとも1種のproteinおよびprotein1の組み合わせであってもよい。 (Embodiment of information provision method)
Next, an embodiment of the information providing method will be described. The embodiment of the information providing method is
-A step of measuring the abundance of at least protein1 among the proteins in the biological sample of the subject,
including. The step of measuring the abundance is not particularly limited as long as the abundance of protein1 can be measured, and may be a combination of at least one proteinin and protein1 selected from the above-mentioned protein2 to 36.
次に、情報提供方法の実施形態について説明する。情報提供方法の実施形態は、
・被検者の生体サンプル中のタンパク質の内、少なくともprotein1の存在量を測定する工程、
を含む。存在量を測定する工程は、少なくともprotein1の存在量を測定できれば特に制限はなく、上記protein2乃至36から選択される少なくとも1種のproteinおよびprotein1の組み合わせであってもよい。 (Embodiment of information provision method)
Next, an embodiment of the information providing method will be described. The embodiment of the information providing method is
-A step of measuring the abundance of at least protein1 among the proteins in the biological sample of the subject,
including. The step of measuring the abundance is not particularly limited as long as the abundance of protein1 can be measured, and may be a combination of at least one proteinin and protein1 selected from the above-mentioned protein2 to 36.
予測方法の実施形態で説明のとおり、特定の種類の被検者タンパク質量は、非小細胞肺がんの化学療法の効果を予測する方法に用いることができる。
As described in the embodiment of the prediction method, a specific type of subject protein amount can be used in a method for predicting the effect of chemotherapy for non-small cell lung cancer.
(情報提供用キットの実施形態)
次に、情報提供用キットの実施形態について説明する。情報提供用キットの実施形態は、被検者の生体サンプル中のタンパク質の内、少なくともprotein1を特異的に認識する抗体、を含む。情報提供用キットは、少なくともprotein1を特異的に認識する抗体を含めば特に制限はなく、上記protein2乃至36から選択される少なくとも1種のproteinを特異的に認識する抗体およびprotein1を特異的に認識する抗体の組み合わせであってもよい。 (Embodiment of information provision kit)
Next, an embodiment of the information providing kit will be described. An embodiment of the information providing kit includes an antibody that specifically recognizes atleast protein 1 among the proteins in the biological sample of the subject. The information providing kit is not particularly limited as long as it contains at least an antibody that specifically recognizes protein 1, and specifically recognizes an antibody that specifically recognizes at least one protein selected from the above proteins 2 to 36 and protein 1. It may be a combination of antibodies to be used.
次に、情報提供用キットの実施形態について説明する。情報提供用キットの実施形態は、被検者の生体サンプル中のタンパク質の内、少なくともprotein1を特異的に認識する抗体、を含む。情報提供用キットは、少なくともprotein1を特異的に認識する抗体を含めば特に制限はなく、上記protein2乃至36から選択される少なくとも1種のproteinを特異的に認識する抗体およびprotein1を特異的に認識する抗体の組み合わせであってもよい。 (Embodiment of information provision kit)
Next, an embodiment of the information providing kit will be described. An embodiment of the information providing kit includes an antibody that specifically recognizes at
上記情報提供方法の実施形態で説明のとおり、特定の種類の被検者タンパク質量は、非小細胞肺がんの化学療法の効果を予測する方法に用いることができる。したがって、特定の種類の被検者タンパク質を特異的に認識する抗体を用い、ELISA(Enzyme-Linked ImmunoSorbent Assay)、免疫組織染色法等、公知の方法を用いて、特定の種類の被検者タンパク質の存在量を測定してもよい。
As described in the embodiment of the above information provision method, a specific type of subject protein amount can be used in a method for predicting the effect of chemotherapy for non-small cell lung cancer. Therefore, using an antibody that specifically recognizes a specific type of subject protein, and using a known method such as ELISA (Enzyme-Linked ImmunoSorbent Assay) or immunohistochemical staining method, a specific type of subject protein is used. You may measure the abundance of.
抗体は、公知の方法で作製すればよい。また、作製した抗体は、公知の方法によりウェル等に結合することで、デバイスの形態で提供してもよい。ELISAまたは免疫組織染色法によるタンパク質の存在量の測定は、公知の方法により行えばよい。情報提供用キットは、抗体に加え、ELISAまたは免疫組織染色法に必要な試薬を含んでもよい。
The antibody may be produced by a known method. Further, the produced antibody may be provided in the form of a device by binding to a well or the like by a known method. The protein abundance may be measured by ELISA or immunohistochemical staining method by a known method. The informative kit may include, in addition to the antibody, the reagents required for ELISA or immunohistochemical staining.
(予測装置、プログラム、記録媒体の実施形態)
図3を参照して、予測装置、プログラム、記録媒体の実施形態について説明する。図3は予測装置の概略を示す図である。予測装置1は、入力部2、予測モデル、必要に応じて閾値を記憶する記憶部3、予測部4、制御部5及びプログラムメモリ6を少なくとも含んでいる。 (Embodiment of predictor, program, recording medium)
An embodiment of a prediction device, a program, and a recording medium will be described with reference to FIG. FIG. 3 is a diagram showing an outline of the prediction device. Theprediction device 1 includes at least an input unit 2, a prediction model, a storage unit 3 for storing a threshold value as needed, a prediction unit 4, a control unit 5, and a program memory 6.
図3を参照して、予測装置、プログラム、記録媒体の実施形態について説明する。図3は予測装置の概略を示す図である。予測装置1は、入力部2、予測モデル、必要に応じて閾値を記憶する記憶部3、予測部4、制御部5及びプログラムメモリ6を少なくとも含んでいる。 (Embodiment of predictor, program, recording medium)
An embodiment of a prediction device, a program, and a recording medium will be described with reference to FIG. FIG. 3 is a diagram showing an outline of the prediction device. The
入力部2は、被検者タンパク質量に関する情報を予測装置1に入力できれば特に制限はなく、キーボード、USB等が挙げられる。また、入力部2はインターネット回線を使用しても良い。例えば、インターネット回線を用いて遠隔地の病院で測定した被検者タンパク質量の情報を予測装置1に送信・入力し、インターネット回線を通じて予測結果を送付することで、遠隔地の病院の被検者に対しても適切な化学療法の効果を予測できる。
The input unit 2 is not particularly limited as long as information regarding the amount of protein of the subject can be input to the prediction device 1, and examples thereof include a keyboard and USB. Further, the input unit 2 may use an internet line. For example, by transmitting and inputting information on the amount of subject protein measured at a remote hospital using an internet line to the prediction device 1 and sending the prediction result via the internet line, the subject at a remote hospital The effect of appropriate chemotherapy can be predicted.
記憶部3には、予測モデル、必要に応じて閾値が記憶されている。予測部4は、入力部2により入力された被検者タンパク質量の情報を記憶部3に記憶されている予測モデルに当てはめることで、被検者の化学療法の効果の予測ができる。プログラムメモリ6には、例えば、図3に示すコンピュータを予測装置1として機能させるためのプログラムが格納されている。このプログラムが制御部5により読み出され実行されることで、入力部2、記憶部3及び予測部4の動作制御が行われる。プログラムは、予めコンピュータに記憶しておいても良いし、記録媒体に予測モデル又は閾値と共に記録され、インストール手段を用いてプログラムメモリ6に格納されるようにしてもよい。
The storage unit 3 stores a prediction model and a threshold value as needed. By applying the information on the amount of protein of the subject input by the input unit 2 to the prediction model stored in the storage unit 3, the prediction unit 4 can predict the effect of the subject's chemotherapy. In the program memory 6, for example, a program for making the computer shown in FIG. 3 function as the prediction device 1 is stored. By reading and executing this program by the control unit 5, the operation control of the input unit 2, the storage unit 3, and the prediction unit 4 is performed. The program may be stored in a computer in advance, or may be recorded on a recording medium together with a prediction model or a threshold value and stored in the program memory 6 by an installation means.
図4は、本出願で開示する予測装置1を用いて、被検者の化学療法の効果を予測するための工程を示す図である。プログラムメモリ6に格納されたプログラムが制御部5に読み出されて実行し、先ず、入力部2により、被検者タンパク質量を入力する(S100)。なお、被検者タンパク質量は、予測装置1と接続している質量分析機等の測定結果を直接入力してもよいし、別途測定した測定値を入力してもよい。次に、入力部2により入力された被検者タンパク質量の情報を、記憶部3に記憶されている予測モデルに当てはめる。予測モデルが判別式の場合は、判別式に被検者タンパク質量を当てはめスコアを算出し、必要に応じて閾値と比較する(S110)。そして、得られた予測結果を表示する(S120)。表示方法は、コンピュータの表示手段に表示してもよいし、紙等にプリントアウトしてもよい。
FIG. 4 is a diagram showing a process for predicting the effect of chemotherapy of a subject using the prediction device 1 disclosed in the present application. The program stored in the program memory 6 is read out by the control unit 5 and executed, and first, the amount of the protein of the subject is input by the input unit 2 (S100). As the amount of protein in the subject, the measurement result of a mass spectrometer or the like connected to the prediction device 1 may be directly input, or the measured value separately measured may be input. Next, the information on the amount of the subject protein input by the input unit 2 is applied to the prediction model stored in the storage unit 3. When the prediction model is a discriminant, the amount of protein of the subject is applied to the discriminant to calculate the score, and the score is compared with the threshold value if necessary (S110). Then, the obtained prediction result is displayed (S120). The display method may be displayed on a display means of a computer, or may be printed out on paper or the like.
以下に実施例を掲げ、本出願で開示する実施形態を具体的に説明するが、この実施例は単に実施形態の説明のためのものである。本出願で開示する発明の範囲を限定したり、あるいは制限することを表すものではない。
Examples are given below to specifically explain the embodiments disclosed in the present application, but these embodiments are merely for the purpose of explaining the embodiments. It does not represent limiting or limiting the scope of the invention disclosed in this application.
<実施例1>
以下の手順により、予測モデルを作成し、化学療法の効果を予測する方法を構築した。 <Example 1>
By the following procedure, a predictive model was created and a method for predicting the effect of chemotherapy was constructed.
以下の手順により、予測モデルを作成し、化学療法の効果を予測する方法を構築した。 <Example 1>
By the following procedure, a predictive model was created and a method for predicting the effect of chemotherapy was constructed.
〔患者〕
予測モデルの作成は、以下に示す登録基準を満たす患者を対象とした。図1に示すとおり、化学療法開始前に249名患者から血液を採取した。
(1)組織診または細胞診で非扁平上皮非小細胞肺がんと診断された症例
(2)根治的放射線照射や外科切除の対象とならない臨床病期IIIA、IIIB、IV期の症例
(3)化学療法未施行の症例(ただし術後補助化学療法施行症例は最終投与から6か月以上の間隔があれば登録可とした)
(4)RECISTの規定で測定可能病変を有する症例(CT又はMRIで最長径がスライス幅の2倍以上かつ10mm以上の病変を有する症例)
(5)年齢が20歳以上の症例
(6)カルボプラチン+ペメトレキセド併用療法に耐えうる主要臓器機能を有する症例
(7)治療薬の投与開始日より3ヶ月以上の生存が期待される症例
(8)告知を受けた研究対象予定患者本人から本研究への参加について文書による同意が得られた症例 〔patient〕
The predictive model was created for patients who met the registration criteria shown below. As shown in FIG. 1, blood was collected from 249 patients prior to the start of chemotherapy.
(1) Cases diagnosed as non-flat epithelial non-small cell lung cancer by histology or cytology (2) Cases of clinical stages IIIA, IIIB, IV that are not subject to radical irradiation or surgical resection (3) Chemotherapy Cases without therapy (However, patients with postoperative adjuvant chemotherapy can be registered if there is an interval of 6 months or more after the last administration).
(4) Cases with measurable lesions according to RECIST (cases with lesions whose longest diameter is at least twice the slice width and 10 mm or more by CT or MRI)
(5) Patients aged 20 years or older (6) Patients with major organ functions that can withstand carboplatin + pemetrexed combination therapy (7) Patients expected to survive for 3 months or longer from the start date of therapeutic drug administration (8) Patients scheduled to be the subject of the study who received the notification: Cases in which written consent was obtained from the patient himself / herself to participate in this study.
予測モデルの作成は、以下に示す登録基準を満たす患者を対象とした。図1に示すとおり、化学療法開始前に249名患者から血液を採取した。
(1)組織診または細胞診で非扁平上皮非小細胞肺がんと診断された症例
(2)根治的放射線照射や外科切除の対象とならない臨床病期IIIA、IIIB、IV期の症例
(3)化学療法未施行の症例(ただし術後補助化学療法施行症例は最終投与から6か月以上の間隔があれば登録可とした)
(4)RECISTの規定で測定可能病変を有する症例(CT又はMRIで最長径がスライス幅の2倍以上かつ10mm以上の病変を有する症例)
(5)年齢が20歳以上の症例
(6)カルボプラチン+ペメトレキセド併用療法に耐えうる主要臓器機能を有する症例
(7)治療薬の投与開始日より3ヶ月以上の生存が期待される症例
(8)告知を受けた研究対象予定患者本人から本研究への参加について文書による同意が得られた症例 〔patient〕
The predictive model was created for patients who met the registration criteria shown below. As shown in FIG. 1, blood was collected from 249 patients prior to the start of chemotherapy.
(1) Cases diagnosed as non-flat epithelial non-small cell lung cancer by histology or cytology (2) Cases of clinical stages IIIA, IIIB, IV that are not subject to radical irradiation or surgical resection (3) Chemotherapy Cases without therapy (However, patients with postoperative adjuvant chemotherapy can be registered if there is an interval of 6 months or more after the last administration).
(4) Cases with measurable lesions according to RECIST (cases with lesions whose longest diameter is at least twice the slice width and 10 mm or more by CT or MRI)
(5) Patients aged 20 years or older (6) Patients with major organ functions that can withstand carboplatin + pemetrexed combination therapy (7) Patients expected to survive for 3 months or longer from the start date of therapeutic drug administration (8) Patients scheduled to be the subject of the study who received the notification: Cases in which written consent was obtained from the patient himself / herself to participate in this study.
〔化学療法〕
図1に示す通り、登録基準を満たさない等の理由を有する5名を除いた244人の患者に対して、カルボプラチンおよびペメトレキセドの併用治療を実施した。カルボプラチンおよびペメトレキセドの投与量・投与期間・維持療法に関しては担当医の判断で実施した。なお、表3に一般的な投与量・投与スケジュール、表4に一般的な維持療法のスケジュールを示す。
〔chemical treatment〕
As shown in FIG. 1, 244 patients excluding 5 patients who did not meet the registration criteria were treated with carboplatin and pemetrexed in combination. The dose, duration, and maintenance therapy of carboplatin and pemetrexed were determined by the doctor in charge. Table 3 shows a general dose / administration schedule, and Table 4 shows a general maintenance therapy schedule.
図1に示す通り、登録基準を満たさない等の理由を有する5名を除いた244人の患者に対して、カルボプラチンおよびペメトレキセドの併用治療を実施した。カルボプラチンおよびペメトレキセドの投与量・投与期間・維持療法に関しては担当医の判断で実施した。なお、表3に一般的な投与量・投与スケジュール、表4に一般的な維持療法のスケジュールを示す。
As shown in FIG. 1, 244 patients excluding 5 patients who did not meet the registration criteria were treated with carboplatin and pemetrexed in combination. The dose, duration, and maintenance therapy of carboplatin and pemetrexed were determined by the doctor in charge. Table 3 shows a general dose / administration schedule, and Table 4 shows a general maintenance therapy schedule.
主治医が定期的(6週間毎)に施行するCT画像に基づき、化学療法の効果を確認した。なお、CT画像は、主治医から独立した内科医および放射線科医によって、RECISTガイドライン version 1.1に基づき、化学療法の効果の有無が確認された。化学療法の効果の有無は、6週間毎のCT画像でがんが小さくなっていれば効果有、がんが大きくなった場合は効果無しとした。症状が変化なし或いは評価できなかった148名の患者を除き、96名を予測モデル作成のためのtraining data用の患者とした。96名の患者の内、化学療法の効果が見られた患者(がんが小さくなった、good(RECISTガイドラインのPR/CR))は59名、効果が見られなかった患者(がんが大きくなった、poor(RECISTガイドラインのPD))は37名であった。
The effect of chemotherapy was confirmed based on CT images performed regularly (every 6 weeks) by the attending physician. The CT images were confirmed by physicians and radiologists independent of the attending physician to have the effect of chemotherapy based on the RECIST guideline version 1.1. The presence or absence of the effect of chemotherapy was considered to be effective if the cancer became smaller on CT images every 6 weeks, and no effect if the cancer became larger. Except for 148 patients whose symptoms did not change or could not be evaluated, 96 patients were selected as training data patients for creating a predictive model. Of the 96 patients, 59 patients showed the effect of chemotherapy (cancer became smaller, good (PR / CR of RECIST guidelines)), and 59 patients did not show the effect (cancer became larger). The number of patients (PD of RECIST guidelines) was 37.
なお、RECISTガイドラインの「CR」、「PR」、「PD」は以下の意味である。
・CR:完全奏効(complete response)。すべての標的病変の消失。
・PR:部分奏効(partial response)。ベースライン長径和と比較して標的病変の最長径の和が30%以上減少。
・PD:進行(progressive disease)。治療開始以降に記録された最小の最長径の和と比較して標的病変の最長径の和が20%以上増加。 In addition, "CR", "PR", and "PD" of the RECIST guideline have the following meanings.
-CR: complete response. Disappearance of all target lesions.
-PR: Partial response. The sum of the longest diameters of the target lesion is reduced by 30% or more compared to the sum of the baseline major diameters.
-PD: Progressive disease. The sum of the longest diameters of the target lesion is increased by 20% or more compared to the sum of the smallest maximum diameters recorded since the start of treatment.
・CR:完全奏効(complete response)。すべての標的病変の消失。
・PR:部分奏効(partial response)。ベースライン長径和と比較して標的病変の最長径の和が30%以上減少。
・PD:進行(progressive disease)。治療開始以降に記録された最小の最長径の和と比較して標的病変の最長径の和が20%以上増加。 In addition, "CR", "PR", and "PD" of the RECIST guideline have the following meanings.
-CR: complete response. Disappearance of all target lesions.
-PR: Partial response. The sum of the longest diameters of the target lesion is reduced by 30% or more compared to the sum of the baseline major diameters.
-PD: Progressive disease. The sum of the longest diameters of the target lesion is increased by 20% or more compared to the sum of the smallest maximum diameters recorded since the start of treatment.
〔血液サンプル中の全タンパク質の存在量の測定〕
training data用の患者の化学療法実施前に採取した血液から、定法により血清(サンプル)を分離した。分離した血清から質量分析器(Sciex社製5600)を用い、発現しているタンパク質を網羅解析した。当該解析したタンパク質の存在量と化学療法の効果に関する情報を関連付けた。 [Measurement of total protein abundance in blood samples]
Serum (sample) was separated by a routine method from blood collected before chemotherapy of patients for training data. The expressed proteins were comprehensively analyzed from the separated sera using a mass spectrometer (5600 manufactured by Siex Co., Ltd.). Information on the abundance of the analyzed protein and the effect of chemotherapy was associated.
training data用の患者の化学療法実施前に採取した血液から、定法により血清(サンプル)を分離した。分離した血清から質量分析器(Sciex社製5600)を用い、発現しているタンパク質を網羅解析した。当該解析したタンパク質の存在量と化学療法の効果に関する情報を関連付けた。 [Measurement of total protein abundance in blood samples]
Serum (sample) was separated by a routine method from blood collected before chemotherapy of patients for training data. The expressed proteins were comprehensively analyzed from the separated sera using a mass spectrometer (5600 manufactured by Siex Co., Ltd.). Information on the abundance of the analyzed protein and the effect of chemotherapy was associated.
〔化学療法の効果予測モデル作成用のタンパク質の決定〕
網羅解析したタンパク質の中から、化学療法がgoodと評価された患者59名、poorと評価された患者37名に共通して存在する537個のタンパク質を予測モデル作成のために用いた。 [Determination of proteins for creating a model for predicting the effect of chemotherapy]
From the comprehensively analyzed proteins, 537 proteins commonly present in 59 patients evaluated as good and 37 patients evaluated as poor were used for creating a predictive model.
網羅解析したタンパク質の中から、化学療法がgoodと評価された患者59名、poorと評価された患者37名に共通して存在する537個のタンパク質を予測モデル作成のために用いた。 [Determination of proteins for creating a model for predicting the effect of chemotherapy]
From the comprehensively analyzed proteins, 537 proteins commonly present in 59 patients evaluated as good and 37 patients evaluated as poor were used for creating a predictive model.
なお、予測モデルの作成に用いたタンパク質は、タンパク質毎に変動(分散)を計算し、median値でサンプル間の中央値を統一した(ノーマライズ)。
For the protein used to create the prediction model, the variation (variance) was calculated for each protein, and the median value was unified between the samples (normalization).
上記537個のタンパク質の内、変動が大きい上位約90%の483個のタンパク質を用いて、図5に示す手順により、予測モデル作成用タンパク質の決定を行った。図5の具体的な手順を以下に示す。
Of the above 537 proteins, 483 proteins in the top 90% with large fluctuations were used, and the protein for creating a predictive model was determined by the procedure shown in FIG. The specific procedure of FIG. 5 is shown below.
(1)Training dataとして選択した96名の患者のタンパク質情報を用いて10分割クロスバリデーション(cross validation,CV)を実施した。症例を10分割し、その内の9分割分をCVtraining data、1分割分をCV validation dataに分けた。training dataを用いて、複数の変数を用いて分類モデルを構築できる重み付け得票分類(Weighted Voting)による分類モデルの作成を行った。作成した分類モデルは、test dataを利用してError rateに基づいた予測性能の評価を行った。
(2)候補タンパク質を一つずつ増やしながら、上記分類モデルの構築を繰り返すことで、候補タンパク質の数(m)が異なるセットを作成した。(1)と(2)の工程を10分割されたデータがCV validation dataとして1回ずつ使用されるように計10回繰り返し、10×m個のモデルを作成した。
(3)更に、これらの工程をn回繰り返すことで、候補タンパク質の数がmであるセットを10×n個作成した。作成した候補タンパク質の数が異なる判別モデルの精度をError rateを指標に評価を行い、最終分類デル作成に適切な候補タンパク質数Mを決定した。候補タンパク質の数がMであるセットを10×n個作成することで、10×M×n個のタンパク質(重複を含む)が得られ、10×n個のモデルの中で最も高い頻度で選択されたタンパク質からM番目までのタンパク質をM個選択し、当該選択したタンパク質を用いて重み付け得票分類に基づく最終分類モデルを構築した。
(4)なお、最終分類モデルとは、10×M×n個の候補タンパク質(重複を含む)の中から選択したタンパク質 M個に基づいて、教師群の全症例を予測できるように重み付け得票分類を用いて作成したモデルを意味する。本解析ではn=10,000で実施した。
(5)構築した最終分類モデル(予測モデル)は、作成に用いた教師群とは別の検証群のデータを用いて検証を行うことで、作成した最終分類モデル(予測モデル)の信頼性の評価をすることができる。
(6)そして、goodとpoorを分類するためには、サンプル中のタンパク質の存在量を測定し、測定した存在量を最終分類モデル(予測モデル)にあてはめ、リスクスコアを算出することで分類すればよい。 (1) Cross validation (CV) was performed using the protein information of 96 patients selected as Training data. The cases were divided into 10 parts, 9 parts of which were divided into CV training data, and 1 division was divided into CV validation data. Using training data, we created a classification model by weighted voting, which allows us to build a classification model using multiple variables. For the created classification model, the prediction performance based on the Error rate was evaluated using test data.
(2) By repeating the construction of the above classification model while increasing the number of candidate proteins one by one, sets having different numbers (m) of candidate proteins were created. The steps (1) and (2) were repeated 10 times in total so that the data divided into 10 was used once as CV validation data, and 10 × m models were created.
(3) Further, by repeating these steps n times, 10 × n sets in which the number of candidate proteins was m were prepared. The accuracy of the discriminant model in which the number of prepared candidate proteins is different was evaluated using the Error rate as an index, and the appropriate number of candidate proteins M for preparing the final classification Dell was determined. By creating 10 × n sets in which the number of candidate proteins is M, 10 × M × n proteins (including duplication) are obtained, and the most frequently selected among the 10 × n models. M proteins from the selected protein to the Mth protein were selected, and the final classification model based on the weighted vote classification was constructed using the selected protein.
(4) The final classification model is a weighted vote classification so that all cases in the teacher group can be predicted based on M proteins selected from 10 × M × n candidate proteins (including duplication). Means a model created using. This analysis was performed with n = 10,000.
(5) The constructed final classification model (prediction model) is verified by using the data of the verification group different from the teacher group used for creation, and the reliability of the created final classification model (prediction model) is improved. Can be evaluated.
(6) Then, in order to classify good and poor, the abundance of protein in the sample is measured, the measured abundance is applied to the final classification model (predictive model), and the risk score is calculated. Just do it.
(2)候補タンパク質を一つずつ増やしながら、上記分類モデルの構築を繰り返すことで、候補タンパク質の数(m)が異なるセットを作成した。(1)と(2)の工程を10分割されたデータがCV validation dataとして1回ずつ使用されるように計10回繰り返し、10×m個のモデルを作成した。
(3)更に、これらの工程をn回繰り返すことで、候補タンパク質の数がmであるセットを10×n個作成した。作成した候補タンパク質の数が異なる判別モデルの精度をError rateを指標に評価を行い、最終分類デル作成に適切な候補タンパク質数Mを決定した。候補タンパク質の数がMであるセットを10×n個作成することで、10×M×n個のタンパク質(重複を含む)が得られ、10×n個のモデルの中で最も高い頻度で選択されたタンパク質からM番目までのタンパク質をM個選択し、当該選択したタンパク質を用いて重み付け得票分類に基づく最終分類モデルを構築した。
(4)なお、最終分類モデルとは、10×M×n個の候補タンパク質(重複を含む)の中から選択したタンパク質 M個に基づいて、教師群の全症例を予測できるように重み付け得票分類を用いて作成したモデルを意味する。本解析ではn=10,000で実施した。
(5)構築した最終分類モデル(予測モデル)は、作成に用いた教師群とは別の検証群のデータを用いて検証を行うことで、作成した最終分類モデル(予測モデル)の信頼性の評価をすることができる。
(6)そして、goodとpoorを分類するためには、サンプル中のタンパク質の存在量を測定し、測定した存在量を最終分類モデル(予測モデル)にあてはめ、リスクスコアを算出することで分類すればよい。 (1) Cross validation (CV) was performed using the protein information of 96 patients selected as Training data. The cases were divided into 10 parts, 9 parts of which were divided into CV training data, and 1 division was divided into CV validation data. Using training data, we created a classification model by weighted voting, which allows us to build a classification model using multiple variables. For the created classification model, the prediction performance based on the Error rate was evaluated using test data.
(2) By repeating the construction of the above classification model while increasing the number of candidate proteins one by one, sets having different numbers (m) of candidate proteins were created. The steps (1) and (2) were repeated 10 times in total so that the data divided into 10 was used once as CV validation data, and 10 × m models were created.
(3) Further, by repeating these steps n times, 10 × n sets in which the number of candidate proteins was m were prepared. The accuracy of the discriminant model in which the number of prepared candidate proteins is different was evaluated using the Error rate as an index, and the appropriate number of candidate proteins M for preparing the final classification Dell was determined. By creating 10 × n sets in which the number of candidate proteins is M, 10 × M × n proteins (including duplication) are obtained, and the most frequently selected among the 10 × n models. M proteins from the selected protein to the Mth protein were selected, and the final classification model based on the weighted vote classification was constructed using the selected protein.
(4) The final classification model is a weighted vote classification so that all cases in the teacher group can be predicted based on M proteins selected from 10 × M × n candidate proteins (including duplication). Means a model created using. This analysis was performed with n = 10,000.
(5) The constructed final classification model (prediction model) is verified by using the data of the verification group different from the teacher group used for creation, and the reliability of the created final classification model (prediction model) is improved. Can be evaluated.
(6) Then, in order to classify good and poor, the abundance of protein in the sample is measured, the measured abundance is applied to the final classification model (predictive model), and the risk score is calculated. Just do it.
以下に、タンパク質を1個~36個に絞り込んだ際に、選択された回数の多いタンパク質を順に示す。
Below, when the proteins are narrowed down to 1 to 36, the proteins with the highest number of selections are shown in order.
上記表5~表16に示すように、タンパク質を1個まで絞り込んだ時のタンパク質1個モデル(以下、単に「1個モデル」と記載することがある。また、M個まで絞り込んだ際には「M個モデル」と記載することがある。)で選択された「protein1」は、2個モデル~36個モデルにおいて全て選択されていた。また、2個モデルの「protein1及び2」も3個モデル~36個モデルの全てで選択され、3個モデルの「protein1乃至3」も4個モデル~36個モデルの全てで選択され、以下同様に、4、5、6、7・・・36まで絞り込んだ時のタンパク質の組合せは、当該組合せより多くのタンパク質を組み合わせた時に選択されていた。以上の結果より、選択回数の順位は異なるものの、タンパク質の組合せを絞り込んだ際に、選択回数が上位のタンパク質には共通性が見られた。
As shown in Tables 5 to 16 above, a one-protein model when the number of proteins is narrowed down to one (hereinafter, may be simply referred to as "one model". Also, when the number of proteins is narrowed down to M, the model may be simply referred to as "one model". "Protein 1" selected in "M model") was selected in all of the 2 to 36 models. In addition, the two models " proteins 1 and 2" are also selected in all of the three to 36 models, and the three models "proteins 1 to 3" are also selected in all of the four to 36 models, and so on. In addition, the protein combination when narrowed down to 4, 5, 6, 7 ... 36 was selected when more proteins were combined than the combination. From the above results, although the order of the number of selections was different, when the combinations of proteins were narrowed down, the proteins with the highest number of selections showed commonality.
図6は、タンパク質を1個~483個まで選択した時のError rateを示すグラフである。なお、Error rateとは、(不正解だった評価サンプル数の例数)/(評価サンプルの全例数)を意味し、Error rateが低いほど好ましい。また、図6に示すError rateは、M(例えば、M=1、M=2、M=3、・・・M=36)毎に作成したn=10,000個のモデルの平均値である。また、タンパク質を1個~483個まで選択したとは、選択回数が多い順に選択したことを意味する。例えば、タンパク質を1個~36個まで選択したとは、それぞれ、上記M個モデルの組み合わせを意味する。
FIG. 6 is a graph showing an error rate when 1 to 483 proteins are selected. In addition, "Error rate" means (the number of examples of the number of evaluation samples that were incorrect) / (the total number of evaluation samples), and the lower the Error rate, the more preferable. Further, the Error rate shown in FIG. 6 is an average value of n = 10,000 models created for each M (for example, M = 1, M = 2, M = 3, ... M = 36). .. Further, selecting 1 to 483 proteins means that the proteins were selected in descending order of the number of selections. For example, selecting 1 to 36 proteins means a combination of the above M models, respectively.
〔予測モデルの作成〕
上記1個モデル~36個モデルに示すように、選択回数が上位のタンパク質に共通性が見られた。したがって、少なくとも「protein1」、更に必要に応じて、protein2乃至36から選択される少なくとも1種とを組み合わせて(例えば、2個モデル~36個モデルに示す組み合わせ。)、予測モデルを作成すればよい。なお、予測の精度を挙げるとの観点からは、例えば、Error rate(不正解だった評価サンプル数の例数/評価サンプルの全例数)が小さい方が好ましい。図6に示すグラフでは、
・1個モデルに示すprotein1の場合のError rateは約15.4%、以下、
・2個モデルに示すprotein1および2の組み合わせの場合は約15.4%、
・3個モデルに示すprotein1乃至3の組み合わせの場合は約15.9%、
・4個モデルに示すprotein1乃至4の組み合わせの場合は約15.7%、
・5個モデルに示すprotein1乃至5の組み合わせの場合は約14.6%、
・6個モデルに示すprotein1乃至6の組み合わせの場合は約13.4%、
・7個モデルに示すprotein1乃至7の組み合わせの場合は約12.5%、
・8個モデルに示すprotein1乃至8の組み合わせの場合は約12.0%、
・9個モデルに示すprotein1乃至9の組み合わせの場合は約10.9%、
・10個モデルに示すprotein1乃至10の組み合わせの場合は約10.2%、
・11個モデルに示すprotein1乃至11の組み合わせの場合は約9.5%、
であった。 [Creating a predictive model]
As shown in the 1-36 model above, there was a commonality among the proteins with the highest number of selections. Therefore, a predictive model may be created by combining at least "protain1" and, if necessary, at least one selected from protein2 to 36 (for example, the combination shown in 2 to 36 models). .. From the viewpoint of improving the accuracy of prediction, for example, it is preferable that the error rate (the number of incorrect evaluation samples / the total number of evaluation samples) is small. In the graph shown in FIG. 6,
-The error rate in the case ofprotein 1 shown in one model is about 15.4%, or less.
・ Approximately 15.4% in the case of the combination of protain 1 and 2 shown in the two models.
・ Approximately 15.9% in the case of the combination ofproducts 1 to 3 shown in the three models.
・ Approximately 15.7% in the case of the combination ofproducts 1 to 4 shown in the 4 models.
・ Approximately 14.6% in the case of the combination ofproducts 1 to 5 shown in the 5 models.
・ Approximately 13.4% in the case of the combination ofproducts 1 to 6 shown in the 6 models.
・ Approximately 12.5% in the case of the combination ofproducts 1 to 7 shown in the 7 models.
・ Approximately 12.0% in the case of the combination ofproducts 1 to 8 shown in the 8 models.
・ Approximately 10.9% in the case of the combination ofproducts 1 to 9 shown in 9 models.
・ Approximately 10.2% in the case of the combination ofproducts 1 to 10 shown in 10 models.
・ Approximately 9.5% in the case of the combination ofproducts 1 to 11 shown in 11 models.
Met.
上記1個モデル~36個モデルに示すように、選択回数が上位のタンパク質に共通性が見られた。したがって、少なくとも「protein1」、更に必要に応じて、protein2乃至36から選択される少なくとも1種とを組み合わせて(例えば、2個モデル~36個モデルに示す組み合わせ。)、予測モデルを作成すればよい。なお、予測の精度を挙げるとの観点からは、例えば、Error rate(不正解だった評価サンプル数の例数/評価サンプルの全例数)が小さい方が好ましい。図6に示すグラフでは、
・1個モデルに示すprotein1の場合のError rateは約15.4%、以下、
・2個モデルに示すprotein1および2の組み合わせの場合は約15.4%、
・3個モデルに示すprotein1乃至3の組み合わせの場合は約15.9%、
・4個モデルに示すprotein1乃至4の組み合わせの場合は約15.7%、
・5個モデルに示すprotein1乃至5の組み合わせの場合は約14.6%、
・6個モデルに示すprotein1乃至6の組み合わせの場合は約13.4%、
・7個モデルに示すprotein1乃至7の組み合わせの場合は約12.5%、
・8個モデルに示すprotein1乃至8の組み合わせの場合は約12.0%、
・9個モデルに示すprotein1乃至9の組み合わせの場合は約10.9%、
・10個モデルに示すprotein1乃至10の組み合わせの場合は約10.2%、
・11個モデルに示すprotein1乃至11の組み合わせの場合は約9.5%、
であった。 [Creating a predictive model]
As shown in the 1-36 model above, there was a commonality among the proteins with the highest number of selections. Therefore, a predictive model may be created by combining at least "protain1" and, if necessary, at least one selected from protein2 to 36 (for example, the combination shown in 2 to 36 models). .. From the viewpoint of improving the accuracy of prediction, for example, it is preferable that the error rate (the number of incorrect evaluation samples / the total number of evaluation samples) is small. In the graph shown in FIG. 6,
-The error rate in the case of
・ Approximately 15.4% in the case of the combination of
・ Approximately 15.9% in the case of the combination of
・ Approximately 15.7% in the case of the combination of
・ Approximately 14.6% in the case of the combination of
・ Approximately 13.4% in the case of the combination of
・ Approximately 12.5% in the case of the combination of
・ Approximately 12.0% in the case of the combination of
・ Approximately 10.9% in the case of the combination of
・ Approximately 10.2% in the case of the combination of
・ Approximately 9.5% in the case of the combination of
Met.
したがって、protein1のみを用いた場合でも比較的高い正答率であったが、例えば、Error rateが10%以下(正答率が90%以上)となる11個モデルに示すprotein1乃至11を少なくとも組み合わせて存在量を測定し、必要に応じてタンパク質の組み合わせ数を多くしてもよい。Error rateは、36個モデルに示す36個のタンパク質(protein1乃至36)の場合に最も小さな値(約4.58%)を示したので、以下の実施例では、36個モデルに示す36個のタンパク質を用いて予測モデルを作成した。作成した予測モデル(判別式)を以下に示す。
Therefore, the correct answer rate was relatively high even when only protein 1 was used, but for example, there are at least a combination of protein 1 to 11 shown in the 11 models in which the Error rate is 10% or less (correct answer rate is 90% or more). The amount may be measured and the number of protein combinations may be increased as needed. The Error rate showed the smallest value (about 4.58%) in the case of the 36 proteins (proteins 1 to 36) shown in the 36 model, so in the following examples, the 36 proteins shown in the 36 model A predictive model was created using proteins. The created prediction model (discriminant) is shown below.
S2Ns1×(protein1のスコア- borders1)+S2Ns2×(protein2のスコア- borders2)+・・・S2Ns35×(protein35のスコア-borders35)+S2Ns36×(protein36のスコア-borders36)
S2Ns1x (score of protein1-borders1) + S2Ns2x (score of protein2-borders2) + ... S2Ns35x (score of protein35-borders35) + S2Ns36x (score of protein36-borders36)
なお、上記予測モデル(判別式)の、“S2Ns1”及び“borders1”とは、下記表17に示すproteion1の“S2Ns”及び“borders”の値である“1.013261825”、“0.931564722”である。“S2Ns2”及び“borders2”・・・は、protein2の“S2Ns”及び“borders”の値を示し、同様に、“S2Ns36”及び“borders36”・・・は、protein36の“S2Ns”及び“borders”の値を示す。また、“proteion1”等の「スコア」とは、測定したタンパク質のノーマライズした後の存在量を用いて算出した判別指数を表す。
上記の予測モデル(判別式)に96サンプルを当てはめることで、各々のサンプルのリスクスコアを算出した。タンパク質が36個以外の予測モデル(判別式)の場合も、同様に計算をすることでリスクスコアを算出できる。閾値は計算したリスクスコに基づき、適宜設定すればよい。例えば、後述する実施例2では閾値を0とし、スコア≧0:good、スコア<0:poorとしているが、他の値であってもよい。 The "S2Ns1" and "borders1" in the above prediction model (discriminant) are "1.013261825" and "0.931564722" which are the values of "S2Ns" and "borders" of production1 shown in Table 17 below. Is. “S2Ns2” and “borders2” ... indicate the values of “S2Ns” and “borders” of platein2, and similarly, “S2Ns36” and “borders36” ... indicate “S2Ns” and “borders” of productin36. Indicates the value of. Further, the "score" such as "proteion 1" represents a discrimination index calculated by using the abundance of the measured protein after normalization.
By applying 96 samples to the above prediction model (discriminant), the risk score of each sample was calculated. In the case of a predictive model (discriminant) other than 36 proteins, the risk score can be calculated by performing the same calculation. The threshold value may be appropriately set based on the calculated risk score. For example, in Example 2 described later, the threshold value is set to 0, the score is ≧ 0: good, and the score is <0: better, but other values may be used.
上記の予測モデル(判別式)に96サンプルを当てはめることで、各々のサンプルのリスクスコアを算出した。タンパク質が36個以外の予測モデル(判別式)の場合も、同様に計算をすることでリスクスコアを算出できる。閾値は計算したリスクスコに基づき、適宜設定すればよい。例えば、後述する実施例2では閾値を0とし、スコア≧0:good、スコア<0:poorとしているが、他の値であってもよい。 The "S2Ns1" and "borders1" in the above prediction model (discriminant) are "1.013261825" and "0.931564722" which are the values of "S2Ns" and "borders" of production1 shown in Table 17 below. Is. “S2Ns2” and “borders2” ... indicate the values of “S2Ns” and “borders” of platein2, and similarly, “S2Ns36” and “borders36” ... indicate “S2Ns” and “borders” of productin36. Indicates the value of. Further, the "score" such as "
By applying 96 samples to the above prediction model (discriminant), the risk score of each sample was calculated. In the case of a predictive model (discriminant) other than 36 proteins, the risk score can be calculated by performing the same calculation. The threshold value may be appropriately set based on the calculated risk score. For example, in Example 2 described later, the threshold value is set to 0, the score is ≧ 0: good, and the score is <0: better, but other values may be used.
作成した予測モデルを用いてTraining dataのgood(N=59)およびpoor(N=37)の予測を行った。正答率(治療効果があると予測)は、good群では93.2%、poor群では0%あったことから、感度(sensitivity)は93.2%、特異性(specificity)は100%で、全体の分類精度(overall classification accuracy)は95.8%であった。また、図7は、「good群患者」vs「poor群患者」のROC曲線を示しており、AUC(area under the curve:濃度曲線下面積)は0.991と非常に高い値であった。
The training data good (N = 59) and power (N = 37) were predicted using the created prediction model. The correct answer rate (predicted to have a therapeutic effect) was 93.2% in the good group and 0% in the poor group, so the sensitivity was 93.2% and the specificity was 100%. The overall classification accuracy (overall specification accuracy) was 95.8%. Further, FIG. 7 shows the ROC curve of the “good group patient” vs. the “poor group patient”, and the AUC (area under the curve: area under the concentration curve) was a very high value of 0.991.
<実施例2>
[検証群との対比]
上記のとおり、表16に示す36個のタンパク質で作成した予測モデルの感度及び特異性が高かったことから、検証群(validation cohort)との対比を行った。 <Example 2>
[Contrast with verification group]
As described above, since the prediction model prepared with the 36 proteins shown in Table 16 had high sensitivity and specificity, a comparison with the validation cohort was performed.
[検証群との対比]
上記のとおり、表16に示す36個のタンパク質で作成した予測モデルの感度及び特異性が高かったことから、検証群(validation cohort)との対比を行った。 <Example 2>
[Contrast with verification group]
As described above, since the prediction model prepared with the 36 proteins shown in Table 16 had high sensitivity and specificity, a comparison with the validation cohort was performed.
検証は、training cohortとは別の患者94人で行った。検証は以下の手順で行った。
(1)検証群の患者から、化学療法を実施する前に血液を採取。
(2)採取した血液から実施例1と同様の手順によりタンパク質の存在量を測定。
(3)実施例1で作成した予測モデルに検証群のタンパク質の存在量を当てはめ、good群(N=57)、poor群(N=37)に分類(予測)。
(4)検証群に実施例1と同様の化学療法を実施し、全生存期間(治療開始日からの生存期間、Overall Survival:OS)、無増悪生存期間(治療開始日からがんが進行せず安定した状態である期間、Progression-Free Survival:PFS)、CT画像による画像診断を行うことで、good群(N=57)、poor群(N=37)の分類(予測)精度を検証した。 Verification was performed on 94 patients separate from the training cohort. The verification was performed by the following procedure.
(1) Blood was collected from patients in the verification group before chemotherapy was performed.
(2) The abundance of protein was measured from the collected blood by the same procedure as in Example 1.
(3) The abundance of proteins in the verification group was applied to the prediction model created in Example 1, and the proteins were classified into a good group (N = 57) and a poor group (N = 37) (prediction).
(4) Chemotherapy similar to Example 1 was performed in the verification group, and the overall survival time (survival period from the treatment start date, Overall Survival: OS) and progression-free survival (cancer progressed from the treatment start date). The classification (prediction) accuracy of the good group (N = 57) and the poor group (N = 37) was verified by performing image diagnosis using the Therapy-Free Survival (PFS) and CT images during the stable state. ..
(1)検証群の患者から、化学療法を実施する前に血液を採取。
(2)採取した血液から実施例1と同様の手順によりタンパク質の存在量を測定。
(3)実施例1で作成した予測モデルに検証群のタンパク質の存在量を当てはめ、good群(N=57)、poor群(N=37)に分類(予測)。
(4)検証群に実施例1と同様の化学療法を実施し、全生存期間(治療開始日からの生存期間、Overall Survival:OS)、無増悪生存期間(治療開始日からがんが進行せず安定した状態である期間、Progression-Free Survival:PFS)、CT画像による画像診断を行うことで、good群(N=57)、poor群(N=37)の分類(予測)精度を検証した。 Verification was performed on 94 patients separate from the training cohort. The verification was performed by the following procedure.
(1) Blood was collected from patients in the verification group before chemotherapy was performed.
(2) The abundance of protein was measured from the collected blood by the same procedure as in Example 1.
(3) The abundance of proteins in the verification group was applied to the prediction model created in Example 1, and the proteins were classified into a good group (N = 57) and a poor group (N = 37) (prediction).
(4) Chemotherapy similar to Example 1 was performed in the verification group, and the overall survival time (survival period from the treatment start date, Overall Survival: OS) and progression-free survival (cancer progressed from the treatment start date). The classification (prediction) accuracy of the good group (N = 57) and the poor group (N = 37) was verified by performing image diagnosis using the Therapy-Free Survival (PFS) and CT images during the stable state. ..
表18に、good群(N=57)、poor群(N=37)の分類された患者の情報を記載する。
Table 18 shows information on the classified patients in the good group (N = 57) and the poor group (N = 37).
〔全生存期間(Overall Survival)〕
図8の折れ線グラフ(GoodおよびPoor)は、カプランマイヤー曲線を表す。図8から明らかなように、good群に分類された患者の全生存期間は、poor群に分類された患者の全生存期間期間より長かった。全生存期間の中間値である50%(縦軸の0.5)は、good群が25.7か月、poor群が4.6か月で、約5.6倍も長かった。 [Overall Survival]
The line graphs (Good and Poor) in FIG. 8 represent Kaplan-Meier curves. As is clear from FIG. 8, the overall survival time of the patients classified in the good group was longer than the overall survival time of the patients classified in the poor group. The median overall survival of 50% (0.5 on the vertical axis) was 25.7 months in the good group and 4.6 months in the poor group, which was about 5.6 times longer.
図8の折れ線グラフ(GoodおよびPoor)は、カプランマイヤー曲線を表す。図8から明らかなように、good群に分類された患者の全生存期間は、poor群に分類された患者の全生存期間期間より長かった。全生存期間の中間値である50%(縦軸の0.5)は、good群が25.7か月、poor群が4.6か月で、約5.6倍も長かった。 [Overall Survival]
The line graphs (Good and Poor) in FIG. 8 represent Kaplan-Meier curves. As is clear from FIG. 8, the overall survival time of the patients classified in the good group was longer than the overall survival time of the patients classified in the poor group. The median overall survival of 50% (0.5 on the vertical axis) was 25.7 months in the good group and 4.6 months in the poor group, which was about 5.6 times longer.
また、以下の表19に示す患者情報、より具体的には、年齢(75歳以上vs75歳未満)、性別(男vs女)、患者の全身状態“Performance Status:PS”(0/1vs2)、喫煙(無しvs過去/現在有り)、EGFR(その他vs陽性)、ステージ(IIIvsIV)、ステージ(IIIvs再発)の情報を考慮してCox回帰分析を行った。なお、分析ソフトには、Rを用いた。表19に示すとおり、ハザード比は0.16、95%信頼区間は0.09-0.30、p値は9.54×10-9であった。
In addition, the patient information shown in Table 19 below, more specifically, age (75 years old or older vs. less than 75 years old), gender (male vs female), general condition of the patient "Performance Status: PS" (0/1 vs 2),. Cox regression analysis was performed considering information on smoking (none vs past / present), EGFR (other vs positive), stage (III vs IV), and stage (III vs recurrence). R was used as the analysis software. As shown in Table 19, the hazard ratio was 0.16, the 95% confidence interval was 0.09-0.30, and the p value was 9.54 × 10 -9 .
〔無増悪生存期間(Progression-Free Survival)〕
図9の折れ線グラフ(GoodおよびPoor)は、カプランマイヤー曲線を表す。図9から明らかなように、good群に分類された患者の無増悪生存期間は、poor群に分類された患者の無増悪生存期間より長かった。無増悪生存期間の中間値である50%(縦軸の0.5)は、good群が6か月、poor群が1.8か月で、約3.3倍も長かった。 [Progression-Free Survival]
The line graphs (Good and Poor) in FIG. 9 represent Kaplan-Meier curves. As is clear from FIG. 9, the progression-free survival of the patients classified in the good group was longer than the progression-free survival of the patients classified in the poor group. The median progression-free survival of 50% (0.5 on the vertical axis) was 6 months in the good group and 1.8 months in the poor group, which was about 3.3 times longer.
図9の折れ線グラフ(GoodおよびPoor)は、カプランマイヤー曲線を表す。図9から明らかなように、good群に分類された患者の無増悪生存期間は、poor群に分類された患者の無増悪生存期間より長かった。無増悪生存期間の中間値である50%(縦軸の0.5)は、good群が6か月、poor群が1.8か月で、約3.3倍も長かった。 [Progression-Free Survival]
The line graphs (Good and Poor) in FIG. 9 represent Kaplan-Meier curves. As is clear from FIG. 9, the progression-free survival of the patients classified in the good group was longer than the progression-free survival of the patients classified in the poor group. The median progression-free survival of 50% (0.5 on the vertical axis) was 6 months in the good group and 1.8 months in the poor group, which was about 3.3 times longer.
また、以下の表20に示す患者情報、より具体的には、年齢(75歳以上vs75歳未満)、性別(男vs女)、患者の全身状態“Performance Status:PS”(0/1vs2)、喫煙(無しvs過去/現在有り)、EDFR(その他vs陽性)、ステージ(IIIvsIV)、ステージ(IIIvs再発)の情報を考慮してCox回帰分析を行った。なお、分析ソフトには、Rを用いた。表20に示すとおり、ハザード比は0.13、95%信頼区間は0.08-0.24、p値は1.67×10-11であった。
In addition, the patient information shown in Table 20 below, more specifically, age (75 years old or older vs. less than 75 years old), gender (male vs female), general condition of the patient "Performance Status: PS" (0/1 vs 2),. Cox regression analysis was performed considering information on smoking (none vs past / present), EDFR (other vs positive), stage (III vs IV), and stage (III vs recurrence). R was used as the analysis software. As shown in Table 20, the hazard ratio was 0.13, the 95% confidence interval was 0.08 to 0.24, and the p value was 1.67 × 10 -11 .
以上の結果より、実施例で作成した予測モデルは非常に優位であることが統計学上から明らかとなった。
From the above results, it became clear from the statistical point of view that the prediction model created in the examples was extremely superior.
次に、検証群の患者のCT画像に基づき、Good群とPoor群に対する化学療法の効果を確認した。なお、化学療法の効果の確認方法は、実施例1の〔化学療法〕と同じである。表28に結果を示す。なお、表21中のResponseは、RECISTガイドライン version 1.1に基づきに基づき、意味は以下の通りである。
・「CR」、「PR」、「PD」は、上記のとおり。
・SD:安定(stable disease。PRとするには腫瘍の縮小が不十分で、かつPDとするには治療開始以降の最小の最長径の和に比して腫瘍の増大が不十分。
・NE:評価不能 Next, the effect of chemotherapy on the Good group and the Poor group was confirmed based on the CT images of the patients in the verification group. The method for confirming the effect of chemotherapy is the same as that of [Chemotherapy] of Example 1. The results are shown in Table 28. The Response in Table 21 is based on the RECIST guideline version 1.1, and has the following meanings.
-"CR", "PR", and "PD" are as described above.
-SD: stable disease. Tumor shrinkage is insufficient for PR, and tumor growth is insufficient for PD compared to the sum of the smallest and longest diameters since the start of treatment.
・ NE: Cannot be evaluated
・「CR」、「PR」、「PD」は、上記のとおり。
・SD:安定(stable disease。PRとするには腫瘍の縮小が不十分で、かつPDとするには治療開始以降の最小の最長径の和に比して腫瘍の増大が不十分。
・NE:評価不能 Next, the effect of chemotherapy on the Good group and the Poor group was confirmed based on the CT images of the patients in the verification group. The method for confirming the effect of chemotherapy is the same as that of [Chemotherapy] of Example 1. The results are shown in Table 28. The Response in Table 21 is based on the RECIST guideline version 1.1, and has the following meanings.
-"CR", "PR", and "PD" are as described above.
-SD: stable disease. Tumor shrinkage is insufficient for PR, and tumor growth is insufficient for PD compared to the sum of the smallest and longest diameters since the start of treatment.
・ NE: Cannot be evaluated
表21から明らかなように、Good群は化学療法により完全・部分奏効(CR、PR)或いは病変の進行が抑えられている(SD)割合が、約89%であった。一方、Poor群では、化学療法により完全・部分奏効(CR、PR)或いは病変の進行が抑えられている(SD)割合は、約49%であった。また、表21を化学療法により明確に治療効果が得られた(CR、PR)との観点で見た場合、19患者の内、18患者がGood群に属していたことから、奏効例の約95%を予測できたと言える。したがって、本出願で開示する予測モデルを用いることで、臨床的にも優位であることを確認した。
As is clear from Table 21, the proportion of complete / partial response (CR, PR) or lesion progression suppressed (SD) by chemotherapy was about 89% in the Good group. On the other hand, in the Poor group, the rate of complete / partial response (CR, PR) or suppression of lesion progression (SD) by chemotherapy was about 49%. In addition, when Table 21 was viewed from the viewpoint that the therapeutic effect was clearly obtained by chemotherapy (CR, PR), 18 of the 19 patients belonged to the Good group. It can be said that 95% could be predicted. Therefore, it was confirmed that it is clinically superior by using the predictive model disclosed in this application.
本出願における開示により、非小細胞肺がんの化学療法の有効性を予め予測できる。したがって、医療機関や大学医学部などの研究機関等における肺がん患者の検査及び研究に有用である。
The disclosure in this application can predict the effectiveness of chemotherapy for non-small cell lung cancer in advance. Therefore, it is useful for examination and research of lung cancer patients in research institutions such as medical institutions and university medical schools.
1…予測装置、2…入力部、3…記憶部、4…予測部、5…制御部、6…プログラムメモリ
1 ... Prediction device, 2 ... Input unit, 3 ... Storage unit, 4 ... Prediction unit, 5 ... Control unit, 6 ... Program memory
Claims (13)
- 非小細胞肺がんの化学療法の効果を予測するための情報を提供する方法であって、
該方法が、
被検者の生体サンプル中のタンパク質の内、少なくともNucleolar protein 58(Accession:Q9Y2X3)の存在量を測定する工程、
を含む、方法。 A method of providing information for predicting the effects of chemotherapy for non-small cell lung cancer.
The method is
A step of measuring the abundance of at least Nucleolar protein 58 (Accession: Q9Y2X3) among the proteins in the biological sample of the subject.
Including, how. - 存在量を測定する工程が、請求項1に記載のタンパク質であるprotein1(Nucleolar protein 58)に加え、以下の表に記載のprotein2乃至36から選択される少なくとも1種以上のタンパク質の存在量を測定する、
請求項1に記載の方法。
The method according to claim 1.
- 存在量を測定する工程が、protein1およびprotein2の存在量を測定する、
請求項2に記載の方法。 The step of measuring the abundance measures the abundance of protein1 and protein2.
The method according to claim 2. - 存在量を測定する工程が、protein1乃至protein3の存在量を測定する、
請求項2に記載の方法。 The step of measuring the abundance measures the abundance of protein1 to protein3.
The method according to claim 2. - 存在量を測定する工程が、protein1乃至protein11の存在量を測定する、
請求項2に記載の方法。 The step of measuring the abundance measures the abundance of protein 1 to protein 11.
The method according to claim 2. - 存在量を測定する工程が、protein1乃至protein36の存在量を測定する、
請求項2に記載の方法。 The step of measuring the abundance measures the abundance of productsin 1 to 36.
The method according to claim 2. - 化学療法が、
(a)カルボプラチンおよびペメトレキセド、
(b)ペメトレキセドおよびシスプラチン、
から選択される併用療法の1種である、
請求項1~6の何れか一項に記載の方法。 Chemotherapy,
(A) Carboplatin and pemetrexed,
(B) Pemetrexed and cisplatin,
It is one of the combination therapies selected from
The method according to any one of claims 1 to 6. - 非小細胞肺がんの化学療法の効果を予測する方法であって、
該方法が、
非小細胞肺がん患者の生体サンプル中で発現している請求項1~7の何れか一項に記載のタンパク質の存在量に基づき予め構築した予測モデルに、被検者の生体サンプル中で発現しているタンパク質の存在量を当てはめる工程と、
予測モデルに当てはめた被検者のタンパク質の存在量から、化学療法の効果を予測する予測工程と、
を含む、方法。 A method of predicting the effects of chemotherapy for non-small cell lung cancer
The method is
A predictive model preliminarily constructed based on the abundance of the protein according to any one of claims 1 to 7 expressed in a biological sample of a non-small cell lung cancer patient is expressed in a biological sample of a subject. The process of applying the abundance of protein in the body and
A prediction process that predicts the effect of chemotherapy from the abundance of protein in the subject applied to the prediction model,
Including, how. - 予め構築した予測モデルが、
非小細胞肺がん患者の生体サンプル中で発現している請求項1~7の何れか一項に記載のタンパク質の存在量を、統計的手段を用いて作成した判別式および閾値であり、
予測工程が、
被検者の生体サンプル中で発現しているタンパク質の存在量を判別式に当てはめスコアを算出し閾値と比較することで、
化学療法の効果あり、または、化学療法の効果無し、を予測する、
請求項8に記載の方法。 The predictive model built in advance
It is a discriminant and a threshold value prepared by using statistical means for the abundance of the protein according to any one of claims 1 to 7 expressed in a biological sample of a non-small cell lung cancer patient.
The prediction process is
By applying the abundance of protein expressed in the biological sample of the subject to the discriminant formula, calculating the score, and comparing it with the threshold value,
Predict whether chemotherapy is effective or not.
The method according to claim 8. - 非小細胞肺がん患者の生体サンプル中で発現している請求項1~7の何れか一項に記載のタンパク質の存在量に基づき予め構築した予測モデルを少なくとも格納した記憶部と、
被検者の生体サンプル中で発現しているタンパク質の存在量を記憶部に記憶された予測モデルに当てはめることで、被検者の化学療法の効果を予測する演算部と、
を含む、非小細胞肺がんの化学療法の効果を予測する予測装置。 A storage unit containing at least a prediction model constructed in advance based on the abundance of the protein according to any one of claims 1 to 7 expressed in a biological sample of a non-small cell lung cancer patient.
By applying the abundance of protein expressed in the subject's biological sample to the prediction model stored in the memory, the calculation unit that predicts the effect of the subject's chemotherapy, and the calculation unit.
Predictors that predict the effects of chemotherapy for non-small cell lung cancer, including. - コンピュータを、請求項10に記載の予測装置として機能させるためのプログラム。 A program for making a computer function as the prediction device according to claim 10.
- 請求項11に記載のプログラムを記録したコンピュータ読み取り可能な記録媒体。 A computer-readable recording medium on which the program according to claim 11 is recorded.
- 被検者の生体サンプル中のタンパク質の内、少なくともNucleolar protein 58(Accession:Q9Y2X3)を特異的に認識する抗体を含む、
非小細胞肺がんの化学療法の効果を予測するための情報提供用キット。 Among the proteins in the biological sample of the subject, the antibody comprising an antibody that specifically recognizes at least Nucleolar protein 58 (Accession: Q9Y2X3).
An informational kit for predicting the effects of chemotherapy for non-small cell lung cancer.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2022526548A JPWO2021241527A1 (en) | 2020-05-25 | 2021-05-24 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2020-090963 | 2020-05-25 | ||
JP2020090963 | 2020-05-25 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2021241527A1 true WO2021241527A1 (en) | 2021-12-02 |
Family
ID=78744779
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2021/019684 WO2021241527A1 (en) | 2020-05-25 | 2021-05-24 | Method for providing information for predicting effect of chemotherapy on non-small cell lung cancer and information provision kit, method for predicting effect of chemotherapy on non-small cell lung cancer, prediction system for predicting effect of chemotherapy on non-small cell lung cancer, and program and recording medium of prediction system |
Country Status (2)
Country | Link |
---|---|
JP (1) | JPWO2021241527A1 (en) |
WO (1) | WO2021241527A1 (en) |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040009481A1 (en) * | 2001-06-11 | 2004-01-15 | Millennium Pharmaceuticals, Inc. | Compositions, kits, and methods for identification, assessment, prevention, and therapy of human prostate cancer |
JP2008263837A (en) * | 2007-04-19 | 2008-11-06 | Univ Nagoya | Method for evaluating efficacy of drug on pulmonary adenocarcinoma cell |
JP2009502115A (en) * | 2005-07-27 | 2009-01-29 | オンコセラピー・サイエンス株式会社 | Diagnostic method for small cell lung cancer |
WO2016121715A1 (en) * | 2015-01-26 | 2016-08-04 | 国立大学法人名古屋大学 | Method for providing information for evaluating prognosis of lung cancer patient, method for predicting prognosis of lung cancer patient, internal standard, antibody, device for predicting prognosis of lung cancer patient, program for prognosis prediction device, and recording medium |
WO2017221744A1 (en) * | 2016-06-24 | 2017-12-28 | 国立大学法人名古屋大学 | METHOD FOR PROVIDING DATA FOR LUNG CANCER TEST, LUNG CANCER TEST METHOD, LUNG CANCER TEST DEVICE, PROGRAM AND RECORDING MEDIUM OF LUNG CANCER TEST DEVICE, AND miRNA ASSAY KIT FOR LUNG CANCER TEST |
WO2020085937A1 (en) * | 2018-10-24 | 2020-04-30 | Общество С Ограниченной Ответственностью "Онкобокс" | Test system for predicting effectiveness of cancer patient treatment using the preparation bevacizumab (avastin) |
JP2020153742A (en) * | 2019-03-19 | 2020-09-24 | 国立大学法人東海国立大学機構 | Method of providing information for assessing response of advanced non-squamous cell lung cancer patient to chemotherapy |
-
2021
- 2021-05-24 WO PCT/JP2021/019684 patent/WO2021241527A1/en active Application Filing
- 2021-05-24 JP JP2022526548A patent/JPWO2021241527A1/ja active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040009481A1 (en) * | 2001-06-11 | 2004-01-15 | Millennium Pharmaceuticals, Inc. | Compositions, kits, and methods for identification, assessment, prevention, and therapy of human prostate cancer |
JP2009502115A (en) * | 2005-07-27 | 2009-01-29 | オンコセラピー・サイエンス株式会社 | Diagnostic method for small cell lung cancer |
JP2008263837A (en) * | 2007-04-19 | 2008-11-06 | Univ Nagoya | Method for evaluating efficacy of drug on pulmonary adenocarcinoma cell |
WO2016121715A1 (en) * | 2015-01-26 | 2016-08-04 | 国立大学法人名古屋大学 | Method for providing information for evaluating prognosis of lung cancer patient, method for predicting prognosis of lung cancer patient, internal standard, antibody, device for predicting prognosis of lung cancer patient, program for prognosis prediction device, and recording medium |
WO2017221744A1 (en) * | 2016-06-24 | 2017-12-28 | 国立大学法人名古屋大学 | METHOD FOR PROVIDING DATA FOR LUNG CANCER TEST, LUNG CANCER TEST METHOD, LUNG CANCER TEST DEVICE, PROGRAM AND RECORDING MEDIUM OF LUNG CANCER TEST DEVICE, AND miRNA ASSAY KIT FOR LUNG CANCER TEST |
WO2020085937A1 (en) * | 2018-10-24 | 2020-04-30 | Общество С Ограниченной Ответственностью "Онкобокс" | Test system for predicting effectiveness of cancer patient treatment using the preparation bevacizumab (avastin) |
JP2020153742A (en) * | 2019-03-19 | 2020-09-24 | 国立大学法人東海国立大学機構 | Method of providing information for assessing response of advanced non-squamous cell lung cancer patient to chemotherapy |
Non-Patent Citations (5)
Title |
---|
AZUMA, YOKO: "SIRT6 expression is associated with poor prognosis and chemosensitivity in patients with non-small cell lung cancer", JOURNAL OF SURGICAL ONCOLOGY, vol. 112, no. 2, 15 July 2015 (2015-07-15), pages 231 - 237, XP055880367 * |
FUKATSU, ASUKI, JAPANESE JOURNAL OF LUNG CANCER, vol. 60, no. 6, 20 October 2020 (2020-10-20), pages 594 * |
ROSELL, R.: "Pemetrexed combination therapy in the treatment of non-small cell lung cancer", SEMINARS IN ONCOLOGY, vol. 29, no. 2, April 2002 (2002-04-01), pages 23 - 29, XP009002524, DOI: 10.1053/sonc.2002.30768 * |
SHI, C. L.: "Notch 3 protein, not its gene polymorphism, is associated with the chemotherapy response and prognosis of advanced NSCLC patients", CELL PHYSIOL BIOCHEM, vol. 34, 19 August 2014 (2014-08-19), pages 743 - 752, XP055880371 * |
SONE, KAZUKI: "CYFRA 21-1 as a predictive marker for non-small cell lung cancer treated with pemetrexed-based chemotherapy", ANTICANCER RESEARCH, vol. 37, no. 2, 2017, pages 935 - 939, XP055880369 * |
Also Published As
Publication number | Publication date |
---|---|
JPWO2021241527A1 (en) | 2021-12-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP7431760B2 (en) | Cancer classifier models, machine learning systems, and how to use them | |
US11621080B2 (en) | Methods and machine learning systems for predicting the likelihood or risk of having cancer | |
Altman et al. | Reporting recommendations for tumor marker prognostic studies (REMARK): explanation and elaboration | |
Adegunsoye et al. | Phenotypic clusters predict outcomes in a longitudinal interstitial lung disease cohort | |
US20200025766A1 (en) | Methods for the detection and treatment of lung cancer | |
CN106714556A (en) | Methods and systems for determining autism spectrum disorder risk | |
Nakagawa et al. | Quantitative CT analysis of honeycombing area in idiopathic pulmonary fibrosis: correlations with pulmonary function tests | |
JP2020522690A (en) | Method and system for identifying or monitoring lung disease | |
CN105793710B (en) | For the composition of pulmonary cancer diagnosis, method and kit | |
KR101378919B1 (en) | System biological method of biomarker selection for diagnosis of lung cancer, subtype of lung cancer, and biomarker selected by the same | |
US20240159753A1 (en) | Methods for the detection and treatment of lung cancer | |
US20230263477A1 (en) | Universal pan cancer classifier models, machine learning systems and methods of use | |
Yutaka et al. | Prognostic impact of preoperative comorbidities in geriatric patients with early-stage lung cancer: Significance of sublobar resection as a compromise procedure | |
Lu et al. | Development and validation of a radiomics nomogram for prognosis prediction of patients with acute paraquat poisoning: a retrospective cohort study | |
Zhu et al. | A Prediction Rule for Overall Survival in Non‐Small‐Cell Lung Cancer Patients with a Pathological Tumor Size Less Than 30 mm | |
JP2020505928A (en) | Method for indicating the presence or absence of prostate cancer in an individual with certain characteristics | |
Sazonova et al. | Development and validation of diffuse idiopathic pulmonary neuroendocrine hyperplasia diagnostic criteria | |
WO2021241527A1 (en) | Method for providing information for predicting effect of chemotherapy on non-small cell lung cancer and information provision kit, method for predicting effect of chemotherapy on non-small cell lung cancer, prediction system for predicting effect of chemotherapy on non-small cell lung cancer, and program and recording medium of prediction system | |
WO2017221744A1 (en) | METHOD FOR PROVIDING DATA FOR LUNG CANCER TEST, LUNG CANCER TEST METHOD, LUNG CANCER TEST DEVICE, PROGRAM AND RECORDING MEDIUM OF LUNG CANCER TEST DEVICE, AND miRNA ASSAY KIT FOR LUNG CANCER TEST | |
Pérez et al. | Rationale and design of the prognostic transcriptomic signature in fibrotic hypersensitivity pneumonitis (PREDICT) study | |
WO2016121715A1 (en) | Method for providing information for evaluating prognosis of lung cancer patient, method for predicting prognosis of lung cancer patient, internal standard, antibody, device for predicting prognosis of lung cancer patient, program for prognosis prediction device, and recording medium | |
Omura et al. | Detection of EGFR mutations in early-stage lung adenocarcinoma by machine learning-based radiomics | |
CN113699235B (en) | Application of immunogenic cell death related gene in head and neck squamous cell carcinoma survival prognosis and radiotherapy responsiveness | |
González Maldonado | Evaluation of Risk Models and Biomarkers for the Optimization of Lung Cancer Screening | |
Wang et al. | The study of correlation between nomogram prediction of uric acid and different chemotherapy regimens in breast cancer patients |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21814454 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2022526548 Country of ref document: JP Kind code of ref document: A |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 21814454 Country of ref document: EP Kind code of ref document: A1 |