Huppert's disease molecule parting and its application on medication guide
Technical field
The invention belongs to biotechnology more particularly to Huppert's disease molecule parting and its on medication guide
Using.
Background technology
Huppert's disease (Multiple Myeloma, MM) is a kind of tumour caused by plasma cell dyscrasias is proliferated,
It is the second common neoplastic hematologic disorder, is 1~2/0 ten thousand people in the incidence of China.Huppert's disease is to be apt to occur in the age to be more than
In 60 years old elderly populations, with the exacerbation of China's aging degree, incidence rises year by year, it has also become seriously threatens old age
A kind of disease of people's health.The characteristic feature of Huppert's disease be marrow in there are the thick liquid cell of a large amount of paraplasms, it is this
Thick liquid cell can secrete a kind of abnormal immunoglobulin or immunoglobulin fragment, i.e. M albumen.
With the application of proteasome inhibitor such as bortezomib and immunoregulation medicament such as lenalidomide, multiple marrow
The Survival of tumor has apparent improvement.But Huppert's disease still can not be cured completely at present.Multiple marrow
Tumor is in the heterogeneity biologically and clinically with height, therefore, to raw after the reaction and treatment of drug combination therapy
Deposit situation has huge difference in different patients.The biological mechanism of this species diversity is caused not yet fully to be managed at present
Solution hinders the personalized progress precisely treated to a certain extent.Therefore, in order to deepen to Huppert's disease biology sheet
It is extremely urgent to develop simple and reliable molecule parting system for the understanding of matter, adjuvant clinical Treatment decsion.Currently, in the world
There are several Huppert's disease molecule parting systems to be suggested.For example, Bergsagel et al. identifies 8 kinds with different
Cyclin D1 (Cyclin D) is expressed and the Huppert's disease hypotype of chromosome translocation.Use unbiased turning without hypothesis
Group analysis is recorded, Zhan and Broyl et al. propose that Huppert's disease has 7-10 molecular isoform, according to the length of patient's life cycle
Short, these hypotypes can be further simplified as high risk group and low-risk group.In addition, with the relevant allelic expression of prognosis
Such as UAMS-70 and UAMS-17, UAMS-80, IFM-15, Millennium-100, EMC-92, gene magnification index such as GPI-5,
MRC-IX-6 and Centrosomal Amplification index are also suggested.But the above molecule parting and expression characteristic can not predict drug therapy
Reaction, can not be associated with the growth course of thick liquid cell, and for molecule parting gene and myelomatosis multiplex because
Between association be not also elucidated with.
Invention content
It is an object of the present invention to provide obtain or detect 97 gene tables in huppert's disease tumour patient to be measured
The purposes of the substance reached.
The present invention provides the substances for obtaining or detecting 97 gene expressions in huppert's disease tumour patient to be measured to exist
Prepare the product for detecting huppert's disease tumour patient bortezomib to be measured or the medication effect containing bortezomib
In application.
The present invention also provides the substances for obtaining or detecting 97 gene expression in huppert's disease tumour patient to be measured
Preparing answering in instructing huppert's disease tumour patient bortezomib to be measured or drug medication product containing bortezomib
With.
Another object of the present invention is to provide 97 gene tables in acquisition or detection huppert's disease tumour patient to be measured
The purposes of the equipment of the substance and operation Huppert's disease Bayes classifier that reach.
The present invention provides obtain or detect in huppert's disease tumour patient to be measured the substance of 97 gene expression and
The equipment for running Huppert's disease Bayes classifier is preparing detection huppert's disease tumour patient to be measured with boron for assistant
Application in the product of rice or medication effect containing bortezomib.
The present invention also provides the substances for obtaining or detecting 97 gene expression in huppert's disease tumour patient to be measured
Equipment with operation Huppert's disease Bayes classifier is preparing guidance huppert's disease tumour patient boron to be measured for assistant
Application in rice or the drug medication product containing bortezomib;
Above-mentioned 97 genes are as follows:
ACBD3、ADAR、ADSS、ALDH2、ANP32E、ANXA2、ATF3、ATP8B2、CACYBP、CAPN2、CCND1、
CCT3、CDC42SE1、CERS2、CHSY3、CLIC1、CLMN、COPA、CSNK1G3、DAP3、DENND1B、ENSA、EPRS、
EPSTI1、EVL、FAM13A、FAM49A、FLAD1、FRZB、GLRX2、HAX1、HDGF、HLA-A、HLA-B、HLA-C、HLA-F、
HLA-G、IL6R、ISG20L2、JTB、KLF2、LAMTOR2、LDHA、MCL1、MOXD1、MRPL24、MRPL9、MVP、MYL6、
NDUFS2、NOP58、NOTCH2NL、NTAN1、PAK1、PI4KB、PIEZO1、PIK3AP1、PIM2、PIP5K1B、PMVK、POGZ、
PPIA、PRCC、PRKCA、PRRC2C、PSMB4、PSMD4、RAB29、RCBTB2、SCAMP3、SCAPER、SDHC、SEL1L3、
SELPLG、SHC1、SIDT1、SSR2、STAP1、TAP1、TIMM17A、TLR10、TMCO1、TOR1AIP2、TOR3A、
TP53INP1、TPM3、TRANK1、TROVE2、UAP1、UBE2Q1、UBQLN4、UHMK1、VPS45、YY1AP1、ZC3H11A、
ZFP36 and ZNF593;
Above-mentioned Huppert's disease Bayes classifier is obtained according to the method included the following steps:
1) the expression quantity data of 97 genes of n Huppert's disease sample are obtained;
2) by the expression quantity data Consensus of 97 genes of the n Huppert's disease sample
Clustering clustering algorithms are divided into two hypotypes of MCL1-M-High and MCL1-M-Low;
3) the expression quantity number of 97 genes of n Huppert's disease sample of two hypotypes based on step 2, step 1)
According to the prognosis life cycle data of, n Huppert's disease sample, built to obtain plain Bayes's classification with Nae Bayesianmethod
Device.
Above-mentioned steps 3) it is that the n Huppert's disease sample is first more than 1 according to sample size ratio:1 ratio with
Machine divides training set and verification collects;The expression quantity data of 97 genes in training are reused, and in conjunction with the use
Each sample MCL1-M-High and MCL1-M-Low hypotype labels that Consensus Clustering clustering algorithms obtain, use R
NB Algorithm in language machine learning packet klaR packets establishes prediction single patient MCL1-M-High hypotypes and MCL1-
The Huppert's disease Bayes classifier of M-Low hypotypes;
The mode of the expression quantity data of 97 genes described above for obtaining each Huppert's disease sample be detection or
Person obtains from database.
Above-mentioned acquisition or the substance for detecting 97 gene expression in huppert's disease tumour patient to be measured are from database
Middle acquisition or the expression quantity for detecting 97 genes in each sample tumour cell (detection method is conventional method).
In the said goods, the n Huppert's disease sample is 551 samples.
Or described it is more than 1:1 ratio is according to 2:1 ratio random division training set and verification collect.
3rd purpose of the invention is to provide a kind of product.
Product provided by the invention, including obtain or detect 97 gene expressions in huppert's disease tumour patient to be measured
Substance and operation Huppert's disease Bayes classifier equipment (equipment can be CD or computer etc..).
In the said goods,
The product has following function:It detects huppert's disease tumour patient bortezomib to be measured or is replaced containing boron
It helps the medication effect of rice or instructs huppert's disease tumour patient bortezomib to be measured or the drug containing bortezomib
Medication.
In the said goods,
The product further includes record 1) or the 2) carrier of detection method;
1) detection method shown in includes the following steps:With the acquisition or detect huppert's disease tumour patient to be measured
In the substances of 97 gene expressions obtain the expression quantity data of described 97 genes of huppert's disease tumour patient to be measured;Again will
The expression quantity data Huppert's disease Bayes's classification of described 97 genes of huppert's disease tumour patient to be measured
Device is classified, and belongs to the huppert's disease tumour patient bortezomib to be measured of MCL1-M-High hypotypes or containing boron for assistant
The medication effect of rice is better than the huppert's disease tumour patient to be measured for belonging to MCL1-M-Low hypotypes;
2) detection method shown in includes the following steps:With the acquisition or detect huppert's disease tumour patient to be measured
In the substances of 97 gene expressions obtain the expression quantity data of described 97 genes of huppert's disease tumour patient to be measured;Again will
The expression quantity data Huppert's disease Bayes's classification of described 97 genes of huppert's disease tumour patient to be measured
Device is classified, if huppert's disease tumour patient to be measured belongs to MCL1-M-High hypotypes, with bortezomib or contain
The drug therapy of bortezomib;If huppert's disease tumour patient to be measured belongs to MCL1-M-Low hypotypes, boron is not had to for assistant
Rice or the drug therapy containing bortezomib.
In the said goods,
The huppert's disease tumour patient to be measured is single patient or multiple patients.
4th purpose of the invention is to provide the method that structure carries out huppert's disease tumour patient on the model of parting.
Method provided by the invention, includes the following steps:
1) the expression quantity data of 97 genes of n Huppert's disease sample are obtained;
2) by the expression quantity data Consensus of 97 genes of the n Huppert's disease sample
Clustering clustering algorithms are divided into two hypotypes of MCL1-M-High and MCL1-M-Low;
3) expression quantity of 97 genes of n Huppert's disease sample of two hypotypes based on step 2), step 1)
Data, the prognosis life cycle data of n Huppert's disease sample, are built to obtain plain Bayes's classification with Nae Bayesianmethod
Device is purpose model.
The present invention also provides a kind of sides judging whether drug of the patient to bortezomib or containing bortezomib be sensitive
Method includes the following steps:
Detect or obtain the expression quantity data of described 97 genes of huppert's disease tumour patient to be measured;It is waited for again by described
The expression quantity data for surveying huppert's disease tumour 97 genes of patient are carried out with above-mentioned Huppert's disease Bayes classifier
Classification, if huppert's disease tumour patient to be measured belongs to MCL1-M-High hypotypes, which suffers from
Medicaments insensitive of the person to bortezomib or containing bortezomib.
The present invention also provides a kind of sides judging whether drug of the patient to bortezomib or containing bortezomib be sensitive
Method includes the following steps:
Detect or obtain the expression quantity data of described 97 genes of huppert's disease tumour patient to be measured;It is waited for again by described
The expression quantity data for surveying huppert's disease tumour 97 genes of patient are carried out with above-mentioned Huppert's disease Bayes classifier
Classification, if huppert's disease tumour patient to be measured belongs to MCL1-M-Low hypotypes, which suffers from
Drug of the person to bortezomib or containing bortezomib is insensitive.
The present invention also provides a kind of methods of guidance huppert's disease tumour patient bortezomib medication to be measured, including
Following steps:Detect or obtain the expression quantity data of described 97 genes of huppert's disease tumour patient to be measured;It again will be described
The expression quantity data of 97 genes of huppert's disease tumour patient to be measured with above-mentioned Huppert's disease Bayes classifier into
Row classification, if huppert's disease tumour patient to be measured belongs to MCL1-M-High hypotypes, with bortezomib or containing boron for assistant
The drug therapy of rice.
The present invention also provides a kind of methods of guidance huppert's disease tumour patient bortezomib medication to be measured, including
Following steps:Detect or obtain the expression quantity data of described 97 genes of huppert's disease tumour patient to be measured;It again will be described
The expression quantity data of 97 genes of huppert's disease tumour patient to be measured with above-mentioned Huppert's disease Bayes classifier into
Row classification does not have to bortezomib or is replaced containing boron if huppert's disease tumour patient to be measured belongs to MCL1-M-Low hypotypes
Help the drug therapy of rice.
The expression quantity of said gene is the gene expression amount in tumour cell.
In order to overcome disadvantages described above, inventor to explore conservative pass during centrum germinativum (GC) plasma cell development
Whether the gene co-expressing network of key signals access can assist illustrating MM pathogenesis and the molecule parting applied to MM.Invention
People's emphasis has found the idiotype network that B cell is controlled in Huppert's disease to imbalance during plasma cell differentiation, because it can
It can play a key effect in the formation of Huppert's disease.By above-mentioned analysis, identifies one and be total to table with MCL1 genes
The netic module (referred to as MCL1-M) reached, and Huppert's disease is divided into two Main Subtypes, i.e. MCL-M- using it
High hypotypes and MCL-M-Low hypotypes.The two hypotypes have dramatically different prognosis and genetics characteristics, it is often more important that,
The categorizing system can also predict treatment of the patient to bortezomib reaction and it is related to the stage of development of thick liquid cell.These hairs
It is existing to have paved road for the implementation that individuation will precisely be treated from now on, can also improve to myelomatosis multiplex because understanding.
Description of the drawings
Fig. 1 is that Bayes classifier genotyping result ROC figures are concentrated in GSE2658 verifications.
Fig. 2 is Bayes classifier genotyping result ROC figures in MMRF data sets.
Fig. 3 is Bayes classifier genotyping result ROC figures in GSE19784 data sets.
Fig. 4 is that the overall survival of Huppert's disease MCL1-M-High and MCL1-M-Low molecular isoform in GSE2658 is bent
Line.
Fig. 5 is that the overall survival of Huppert's disease MCL1-M-High and MCL1-M-Low molecular isoform in GSE2658 is bent
Line.
Fig. 6 is the overall survival of Huppert's disease MCL1-M-High and MCL1-M-Low molecular isoform in GSE19784
Curve (above) and Progression free survival curve (figure below).
Fig. 7 has bortezomib for treating for MCL1-M-High and MCL1-M-Low hypotypes patient in GS19784 different
Reaction.
Specific implementation mode
Experimental method used in following embodiments is conventional method unless otherwise specified.
The materials, reagents and the like used in the following examples is commercially available unless otherwise specified.
The screening of the molecular diagnostic markers of embodiment 1, Huppert's disease and the implementation of molecule parting
The Huppert's disease expression quantity data set GSE2658 provided using NCBI GEO public databases, passes through Pierre
Gloomy correlation analysis obtains 87 genes co-expressed with MCL1, and based on this, identifies 46 in low expression
The gene being enriched in the Huppert's disease sample of MCL1-M genes.For more stable carry out molecule parting, this 133 genes
In 36 classification effect is not high is further screened out, final 97 stable differential expressions and the relatively high classification gene of abundance
It has been retained.
The title of 97 genes is as follows:
ACBD3、ADAR、ADSS、ALDH2、ANP32E、ANXA2、ATF3、ATP8B2、CACYBP、CAPN2、CCND1、
CCT3、CDC42SE1、CERS2、CHSY3、CLIC1、CLMN、COPA、CSNK1G3、DAP3、DENND1B、ENSA、EPRS、
EPSTI1、EVL、FAM13A、FAM49A、FLAD1、FRZB、GLRX2、HAX1、HDGF、HLA-A、HLA-B、HLA-C、HLA-F、
HLA-G、IL6R、ISG20L2、JTB、KLF2、LAMTOR2、LDHA、MCL1、MOXD1、MRPL24、MRPL9、MVP、MYL6、
NDUFS2、NOP58、NOTCH2NL、NTAN1、PAK1、PI4KB、PIEZO1、PIK3AP1、PIM2、PIP5K1B、PMVK、POGZ、
PPIA、PRCC、PRKCA、PRRC2C、PSMB4、PSMD4、RAB29、RCBTB2、SCAMP3、SCAPER、SDHC、SEL1L3、
SELPLG、SHC1、SIDT1、SSR2、STAP1、TAP1、TIMM17A、TLR10、TMCO1、TOR1AIP2、TOR3A、
TP53INP1、TPM3、TRANK1、TROVE2、UAP1、UBE2Q1、UBQLN4、UHMK1、VPS45、YY1AP1、ZC3H11A、
ZFP36 and ZNF593.
This 97 genes then will be as the classified adaptive factor for parting.It is multiple using 551 of GSE2658 data sets
The expression quantity data of this 97 genes in myeloma sample, first using Consensus Clustering clustering algorithms with no prison
This 551 Huppert's diseases have been divided into two hypotypes of MCL1-M-High and MCL1-M-Low by the cluster mode superintended and directed.But it is based on
The sorting technique of cluster cannot be directed to independent sample and carry out molecule parting.In order to implement Individual Diagnosis, this 551 samples according to
2:1 ratio is by random division training set (369) and verification collection (182), the grader for establishing and assessing individuation.
Sampling takes the mode of Stratified Sampling, to ensure two Asias MCL1-M-High and MCL1-M-Low in training set and test set
The ratio of type remains consistent with original.
According to the expression quantity data and Consensus Clustering of 97 classification genes of 369 samples in training set
Two hypotype labels of MCL1-M-High and MCL1-M-Low that clustering algorithm is divided into, use machine learning packet klaR packets in R language
The NB Algorithm of offer establishes the multiple of predictable single patient MCL1-M-High hypotypes and MCL1-M-Low hypotypes
Property myeloma Bayes classifier.
And the accuracy of its classification of 182 Samples Estimates is concentrated using verification.
According to the accuracy of return, continuous iteration optimization model, it has been more than 95% finally to make the accuracy of classification, the classification
The accuracy data of device is shown in Table 1, and receiver operating curves (ROC) see Fig. 1.
The accuracy that table 1. is concentrated using the grader that GSE2658 data sets are established in verification
Grader in order to determine the use of the foundation of GSE2658 data sets expanded can be applied.Inventor is followed by this
Grader predicts Huppert's disease large data collection MMRF and GEO the Huppert's disease expression quantity data set of NCI publications
The molecular isoform of sample in GSE19784.
MMRF data sets are different from GSE2658, and the expression quantity of gene is obtained by RNA-seq rather than chip.It is predicted
Result such as table 2, ROC figures are shown in Fig. 2.
The accuracy in MMRF data sets for the grader that table 2. is established using GSE2658 data sets
The results show that even if cross-platform grader if can keep very high accuracy, this illustrates that it has higher push away
Wide application value.
GSE19784 is also the expression quantity data set of a Huppert's disease, as GSE2618, using U133
2.0 Plus chips measure gene expression amount.But the two is detected by different experiments in different times, tests item
Part may have difference, this causes the data of the two dramatically different in distribution and noise level.The database subtype prediction knot
Fruit is shown in Table 3, and ROC curve is shown in Fig. 3.
The accuracy in GSE19784 data sets for the grader that table 3. is established using GSE2658 data sets
The results show that the grader can preferably overcome the above problem, higher accuracy is still kept.
Application of the Bayes classifier in predicting patient's prognosis survival rate of embodiment 2, Huppert's disease
One, database GSE2658
According to 97 classification genes of 551 multiple myeloma patients samples of GSE2658 databases (detection before treatment)
Expression quantity data, using embodiment 1 obtain Huppert's disease Bayes classifier by 551 sample classifications, obtain
249 MCL1-M-High hypotypes Huppert's diseases and 302 MCL1-M-Low hypotype Huppert's diseases.
Track after follow-up 551 sample patients treatment 72 months, according to Follow-up results, carry out survival analysis (K-M analyze and
Cox regression analyses), the results are shown in Figure 4, it can be seen that the two Huppert's disease Asias MCL1-M-High and MCL1-M-Low
Type has dramatically different prognosis, the low (log-rank of overall survival ratio MCL1-M-Low hypotypes of MCL1-M-High hypotypes
It examines, p=0.0201, likelihood ratio test, Hazard ratio 1.588, p=0.0212).
Therefore, parting is carried out using 97 genes of MCL1 gene groups using Bayes classifier, can be used for predicting to be measured
The prognosis of patient.
Two, database MMRF
According to 534 97 points of MCL1 gene groups of multiple myeloma patients sample (detection before treatment) in MMRF data sets
The Huppert's disease Bayes classifier of the acquisition of embodiment 1 is respectively adopted by 534 samples in the expression quantity data of genoid
It is divided into two hypotypes of MCL1-M-High (231) and MCL1-M-Low (303).
Track after follow-up 534 sample patients treatment 48 months, according to Follow-up results, carry out survival analysis (K-M analyze and
Cox regression analyses), the results are shown in Figure 5, it can be seen that in MMRF data sets, MCL1-M-High and MCL1-M-Low two
A hypotype equally has dramatically different prognosis, the overall survival ratio MCL1-M-Low hypotypes of MCL1-M-High hypotypes low
(log-rank is examined, p=0.006663, likelihood ratio test, Hazard ratio 1.838, p=0.00706).
Should the result shows that, no matter which platform gene expression amount data come from, MCL1 is utilized using Bayes classifier
97 genes of gene group carry out parting, can be used for predicting the prognosis of patient to be measured.
Three, database GSE19784
Collect the MCL1 of 304 multiple myeloma patients samples (detection before treatment) according to verification in database GSE19784
The expression quantity data of 97 classification genes of gene group, are respectively adopted the Huppert's disease Bayes classifier of the acquisition of embodiment 1
304 samples are divided into two hypotypes of MCL1-M-High (107) and MCL1-M-Low (196).
Track after follow-up 304 sample patients treatment 96 months, according to Follow-up results, carry out survival analysis (K-M analyze and
Cox regression analyses), the results are shown in Figure 6 (A is overall survival, and B is Progression free survival rate), it can be seen that in GSE19784
In data set, two hypotypes of MCL1-M-High and MCL1-M-Low equally have dramatically different prognosis, MCL1-M-High sub-
Overall survival ratio MCL1-M-Low hypotypes low (log-rank inspections, the p of type<0.0001, likelihood ratio test, Hazard ratio
1.91, p=0.0002).GSE19784 data sets also include the progress information of disease, therefore we also analyze Progression free survival
The difference of rate, similar, the Progression free survival rate also log-rank inspections lower than MCL1-M-Low hypotype of MCL1-M-High hypotypes
Test, p=0.0282, likelihood ratio test, Hazard ratio 1.36, p=0.031) result again shows that, using Bayes classifier profit
Parting is carried out with 97 genes of MCL1 gene groups, can be used for predicting the prognosis of patient to be measured.
Embodiment 3, the molecular diagnostic markers of Huppert's disease and parting are predicting whether patient to be measured can use boron
Bortezomib is treated
The expression quantity data set of GSE19784 Huppert's diseases comes from the clinical test of an III phase drug
(HOVON-65/GMMG-HD4), has the therapeutic scheme of patient.Patient has been assigned to two groups and received respectively by the experiment at random
Two kinds of pharmaceutical compositions of VAD (155) and PAD (148), the two the difference is that the more bortezomib (trade names of PAD schemes:
Bortezomib) this drug.Enter group patient gene expression amount data be all collected in treatment before.
Patient is layered by above-mentioned MCL1-M molecule partings, then respectively MCL1-M-High (PAD 51,
VAD 56) and two hypotypes of MCL1-M-Low (PAD 104, VAD 92) in carried out into grouping by therapeutic scheme
Survival analysis (K-M is analyzed and cox regression analyses).
The results are shown in Figure 7, and A is the overall survival of MCL1-M-High groups, and B is the overall survival of MCL1-M-Low groups
Rate, C are the Progression free survival rate of MCL1-M-High groups, and D is the Progression free survival rate of MCL1-M-Low groups;It is observed that making
It is only capable of extending the life cycle of patient in MCL1-M-High groups with the PAD drugs of bortezomib, especially progression free survival phase (Fig. 7
Left side, MCL-M-High groups, right side MCL-M-Low groups;Top, overall survival curve, lower section, Progression free survival curve), this takes off
Having shown bortezomib clinically can delay the recurrence of MCL-M-High group patients to deteriorate, but in MCL-M-Low group patients
But without any effect.In conclusion clinical application can be instructed by implementing the molecule parting of the present invention, it can be to avoid in MCL1-
Bortezomib is used in M-Low group patients, on the one hand can mitigate the financial burden of patient, on the one hand patient can also lacked and be undertaken
The side effect that drug therapy is brought.