1342393 (1) 玖、發明說明 【發明所屬之技術領域】 本發明係關於多成份藥劑之評估方法，更詳細而言， 係關於將構成多成份藥劑之各成份的相異程度，對於由三 維高效能液相層析（以下簡稱爲「3D - HPLC」）所測定 之數據，由施以特定的分析步驟，使用所得數値之多成份 藥劑之評估方法者。 【先前技術】 多成份藥劑，尤其漢方製劑等之來自天然物之製劑之 定量的及定性的評估係關於使用原料生藥之地質學因素、 生態學因素、收集時期、收集場所' 收集年代及生育期之 天候等之原因而變化。因此，關於漢方製劑等之多成份藥 劑，爲評估其品質、安全性及效力等，而規定一定的判定 基準，基於該基準，國家之監督機關、化學組織及製造業 者等進行評估。 然而，多成份藥劑品質等之判定基準，一般係適當地 選擇1種至多種之多成份藥劑中之某特徵成份，基於該含 量等而作成。 例如於月刊藥事νο1·28，No.3，67-71(1986)中記載， 關於多成份藥劑，不能鑑定本質成份時，選擇具有可定量 分析，容易溶解於水，於熱水中不分解，不與其他成份化 學反應等之物性之多數個成份，化學分析此等成份之含量 ，以所得之數値爲評估基準。 (2) (2)1342393 另一方面，亦已知測定多成份藥劑之色譜法，得到各 溶離時間之紫外光可見光吸收光譜（以下簡稱爲「指紋圖 譜數據」），由該成份資訊作成評估的基準。例如於特開 20 02 — 2 M2 15號公報中記載，由指紋圖譜數據中選擇幾個 波峰，依條碼化而評估多成份藥劑。 然而，上述方法中，有「特定成份之含量」或「特定 成份之色譜波峰」之槪念，爲進行其定量，於色譜記錄表 上，將特定成份之色譜波峰與其他成份之色譜波峰分離， 或爲求特定成份之波峰面積或高度，將需要由電子計算機 之波形處理操作，此事成爲降低數據正確性的原因之一》 亦即，波峰之溶離時間並無於嚴格意義下之再現性，因此 ，波峰之波形處理發生分散岐（dispersion)。尤其，此 分岐於波形處理小廣底波峰或連續波峰時更爲顯著，缺乏 作爲評估方法之信賴性者。另外，由電子計算機進行波形 處理時，一般需要龐大的時間。 另外，上述方法中，因爲資訊量（數據點量）局限於 特定成份之波峰數，無法自由地增減資訊量，也因爲無法 調整處理數據時間，無法確立最適合的評估方法。 另外*上述方法中，因爲多數個特定成份含量作爲多 數個數値而得到，必須綜合此等數値而判斷，不能以一個 數値評估多成份藥劑。亦即，需評估之一個多成份藥劑， 係不能以一個數値表示與多數個多成份藥劑群有多少差異 程度。 在此，要求分岐少，信賴性高，亦改善處理數據時間 -6- (3) (3)1342393 ’並且可以一個數値判定之多成份藥劑之評估方法。 【發明內容】 發明之揭示 本發明者等係有鑑於上述情況而努力檢討的結果，發 現使用3D — HPLC之指紋圖譜數據，以特定的方法處理此 指紋圖譜數據之各項目，進行數理分析，以一個數値而可 評估多成份藥劑之品質、安全性及效力等之方法，而完成 本發明。 亦即，本發明係提供以使用至少由步驟（1 )至（5 ) (1) 得到需評估之多成份藥劑之3D - HP LC之指紋圖譜 數據之步驟 (2) 將上述（1 )所得之指紋圖譜數據，與構成基準群 之其他同種類多成份藥劑之3D - HPLC之指紋圖譜數據組 合之步驟 (3) 關於上述（2 )之指紋圖譜數據，對於多成份藥劑 之號碼，與任一項之溶離時間或檢測波長，分配爲MT法 (田口 品質工程法，Mahalan obis Taguchi Method )中之 變數軸，以訊號強度爲MT法中之特徵量之步驟 (4) 由上述（3 )之特徵量，使用Μ T法而得到單位空 間之步驟 (5) 由上述（4 )所得之單位空間，使用ΜΤ法而得到 需評估之多成份藥劑之馬氏距離之步驟 所得之馬氏距離（Mahalanobis1 distance)，判定上 (4) (4)1342393 述需評估之多成份藥劑與作爲基準群所選擇之多數個多成 份藥劑之相異程度爲特徵之多成份藥劑之評估方法者。 另外，本發明係提供以使用至少由步驟（1 )至（8 ) (1) 得到需評估之多成份藥劑之3D_ HP LC之指紋圖譜 數據之步驟 (2) 將上述（1 )所得之指紋圖譜數據，與構成基準群 之其他同種類多成份藥劑之3 D - HPLC之指紋圖譜數據組 合之步驟 (3) 關於上述（2 )之指紋圖譜數據，對於多成份藥劑 之號碼，與任一項之溶離時間或檢測波長，分配爲MT法 (磁粒檢測法）中之變數軸，以訊號強度爲MT法中之特 徵量之步驟 (4) 由上述（3)之特徵量，使用MT法而得到單位空 間之步驟 (5 )由上述（4 )所得之單位空間，於各檢測波長或溶 離時間，使用MT法而得到全部多成份藥劑之馬氏距離之 步驟 (6) 對於多成份藥劑的號碼與步驟（3)之未分配爲變 數軸之另一項，分配爲MT法中之變數軸，以步驟（5) 所得之馬氏距離爲MT法中之特徵量之步驟 (7) 由上述（6 )之特徵量，使用MT法而得到第2單位 空間之步驟 (8) 由上述（7 )所得之第2單位空間，使用MT法而得 到需評估之多成份藥劑之馬氏距離之步驟 (5) (5)1342393 所得之馬氏距離，判定上述需評估之多成份藥劑與作 爲基準群所選擇之多數個多成份藥劑之相異程度爲特徵之 多成份藥劑之評估方法。 另外，本發明係提供實施上述評估方法用之程式及/ 或記錄得到單位空間用之基準群指紋圖譜數據之記錄媒體 者。 用以實施發明之最佳型態 本發明係將指紋圖譜數據，由使用MT法數理分析所 得之馬氏距離，評估多成份藥劑之方法。 本發明方法中係有包含步驟（1 )至（5 )之進行1次 分配變數軸之方法（以下稱爲「本發明第1方法」），與 包含步驟（1 )至（8 )之進行2次分配變數軸之方法（以 下稱爲「本發明第2方法」）。 實施本發明第1方法係首先作爲步驟（1 )，必須得到 多成份藥劑之3D - HP LC之指紋圖譜數據。 在此’所謂多成份藥劑係定義爲含有多數個有效化學 成份之藥劑，雖無特別限定者，但適宜的是例如植物萃取 物、生藥或使用此等之漢方製劑等。另外，劑形亦無特別 地限制’包括例如湯劑、漢方萃取顆粒、漢方萃取液、漢 方錠劑及漢方膠囊劑等。 本發明中所使用之3D_ HPLC係指於各個所定之多數 個溶離時間所得之溶出成份之光譜。 作爲此3D — HP LC所使用之管柱，通常係使用高效液 (6) (6)1342393 相層析所使用者。 另外，3D — HP LC之檢測波長，雖無特別的限制，但 以150nm至900nm之範圍爲宜，以200nm至400nm之紫外 線可見光吸收領域尤佳，以選自200nm至300nm之多數個 波長更好。 關於光譜之訊號強度，透過率（transmittance)或吸 光度（absorance)皆可，但以吸光度爲宜。 所謂步驟（1 )所得之指紋圖譜，係指使用上述3D -HPLC，例如對於溶離時間與檢測波長，上述光譜之訊號 強度係以3維所表示者，例如圖丨（a )所示者。因此，所 謂指紋圖譜數據係至少具有多成份藥劑之號碼（批號）、 溶離時間、檢測波長及訊號強度爲數據。 使用3D _HPLC之指紋圖譜數據亦可由市售裝置而取 得，作爲相關的市售裝置，可舉例如島津製作所製之LC —VP系統等》 本發明第1方法中之步驟（2)係將上述（1)所得之 指紋圖譜數據，與構成基準群之其他同種類多成份藥劑之 3D - HPLC之指紋圖譜數據組合之步驟。 構成基準群之其他同種類多成份藥劑之3D- HPLC之 指紋圖譜數據亦可以上述方法取得。測定條件等之差異並 無特別地限定，但以需評估之多成份藥劑與構成基準群之 其他同種類多成份藥劑亦使用相同的管柱’於相同條件下 測定爲宜。 所謂構成基準群之其他同種類多成份藥劑係指例如漢 -10- (7) 1342393 方製劑時，即使以相同的名稱稱呼者，因 場所、收集年代、生育期之天候等相異， 生態學因素及氣候學因素等，所以其成份 即使此等因素不同亦可之多數個之多成份 基準群雖選自多數個多成份藥劑，但 製造之多成份藥劑群爲宜。另外，以得到 全部的多成份藥劑群爲宜。 另外，明白成爲標準之地質學因素、 候學因素等時，則以選擇包含於該因素範 爲基準群爲宜。例如可限定採取年，或採 以選擇一定期間內，由特定公司所銷售之 。另外，尤其具有一定的藥效，使成爲基 份藥劑時，以選擇此等爲宜。另外，選擇 同一名稱所銷售等之全部的多成份藥劑爲 估之一個多成份藥劑本身亦可使用於構成 構成基準群之多成份藥劑之數量並無 5個以上爲宜，以100個以上尤佳》 本發明第1方法中之步驟（3 )係上述 ，對於多成份藥劑之號碼，與任一項之溶 長，分配爲MT法中之變數軸，以訊號強 特徵量之步驟。 在此本發明之MT法係指現在品質工 —般已知之計算手法者，舉例如「品質工 規格協會發行（2 0 00 )第1 36至1 38頁、品 收集時期、收集 依地質學因素、 組成發生改變’ 藥劑。 以某一家公司所 指紋圖譜數據之 生態學因素及氣 圍之多成份藥劑 取場所。另外， 多成份藥劑爲宜 準之多數個多成 由多數個公司以 宜。另外，需評 基準群用。 特別的限制，以 指紋圖譜數據中 離時間或檢測波 度爲MT法中之 業上，MT法爲 學的數理」日本 質工學應用講座 -11 - (8) (8)1342393 「化學、藥學 '生物學之技術開發」日本規格協會編（ 1999)第454至456頁及品質工學，丄丄（5) ，78_84( 2003年）所記載之方法。 另外，亦可使用一般市售之MT法程式軟體。作爲市 售之MT法程式軟體，可舉例如（股）Probe社之PRAT for Research V1.0，PRAT for Development V].0 ;(財） 曰本規格協會之 TM — ANOVA ;(股）Ohken之 MTS V e r. 2 · 0 f o r E x c e 1 及 Μ T f o r w i n d o w s 等。 本發明第1方法中，必須至少對於多成份藥劑之號碼 ，與任一項之溶離時間或檢測波長’分配爲MT法中之變 數軸，以訊號強度爲MT法中之特徵量。 雖然對於變數軸之分配並無特別的限定’但以分配溶 離時間於Μ T法中所謂的項目軸（X軸）’分配多成份藥 劑之號碼於所謂號碼列軸（y軸）’分配訊號強度於mt 法中所謂的特徵量爲宜。 在此，上述項目軸（X軸）與號碼列軸（y軸）係如 下定義。亦即於MT法中’關於下述表1中之數據組xi】’ 求出平均値mj與標準偏差σ j ’由規格化Χί·ΐ値之xij== (X i j - m j ) / σ j，求出i與j之相關係數r ’而得單位空 間或馬氏距離，此時項目軸（x軸）與號碼列軸（y軸） 係如「平均値mj與標準偏差<7 j係於各項目軸（x軸）之 値，改變號碼列軸（y軸）之値而求出」所定義。 (9) 1342393 表1 項目軸（X軸） 號碼列軸（Y軸） 項目1項目2 ···項目j項目k 多成份藥劑之號碼1 多成份藥劑之號碼2 多成份藥劑之號碼i 多成份藥劑之號碼η X ! 1 X 1 2 .·· … X , k X2I X22 ··· … X 2 k … ·. ··· X j j … Xnl Xn2 ·" ·* Xnk 本發明第1方法中之步驟（4)係由步驟（3)中之分 配於軸之數據與特徵量，使用Μ T法，得到基準點與單位 量（以下簡稱爲「單位空間」）之步驟。在此，基準點、 單位量及單位空間係依據上述Μ Τ法之文獻之記載所定義 〇 本發明第丨方法中之步驟（5 )係使用ΜΤ法得到需評 估之多成份藥劑之由上述單位空間之馬氏距離（ΜΤ法中 記載爲之D2 )之步驟。在此之馬氏距離（D2 )係與上述 ΜΤ法之文獻說明同樣地定義，另外，馬氏距離係以上述 文獻所記載的方法而求出。 選擇多成份藥劑之號碼與溶離時間作爲變數軸時，得 到於各檢測波長之各多成份藥劑之馬氏距離，選擇多成份 藥劑之號碼與檢測波長作爲變數軸時，得到於各溶離時間 之各多成份藥劑之馬氏距離。例如分配溶離時間於項目軸 (X軸），分配多成份藥劑之號碼於號碼列軸（y軸）| 分配訊號強度於MT法中之特徵量時，得到各檢測波長之 -13- (10) (10)1342393 馬氏距離（D2 )。 使用如此所得之馬氏距離，可適當地判定需評估多成 份藥劑與作爲基準群所選擇之多數個多成份藥劑之相異程 度。 另外，本發明第2方法係適用適用一般稱爲分割合成 法、複合MT法或多階MT法之方法。 實施本發明第2方法時，步驟（1 )至步驟（5 )係與 本發明第1方法同樣地實施。 另外，本發明第2方法中之步驟（6)係對於多成份藥 劑的號碼與步驟（3)中之未分配爲變數軸之另一項，分 配爲MT法中之變數軸，以馬氏距離爲MT法中之特徵量 之步驟。 舉例適合的方法如於步驟（3 )中，分配溶離時間於 Μ T法中所謂的項目軸（X軸），分配多成份藥劑之號碼 於所謂號碼列軸（y軸），分配訊號強度於MT法中所謂 的特徵量時，於各檢測波長，得到對於各多成份藥劑之號 碼之馬氏距離，其次，分配檢測波長於項目軸（X軸）， 分配多成份藥劑之號碼於所謂號碼列軸（y軸）*分配上 述步驟（1 )至步驟（5 )所得之馬氏距離於MT法中所謂 的特徵量的方法。 ' 本發明第2方法中之步驟（7 )係與上述步驟（4 )同 樣地，基於嶄新地分配於軸之數據群，而得到第2單位空 間之步驟。 本發明第2方法中之步驟（8 )係由上述（7 )所得之 -14- (11) (11)1342393 第2單位空間，使用MT法而得到需評估之多成份藥劑之 馬氏距離之步驟》 使用步驟（8 )所得之馬氏距離（以下稱爲「合成馬 氏距離」）’可適當地判定上述一個需評估多成份藥劑與 作爲基準群所選擇之多數個多成份藥劑之相異程度。 使用合成馬氏距離時’因爲可考量全部之溶離時間、 檢測波長及多成份藥劑的號碼而評估，活用3 D — η p L C之 特點’所以特別適宜。 作爲本發明之適合型態，預先由特定的公司以同一名 稱所製造或銷售之多成份藥劑構成基準群，得到3 D -HPLC指紋圖譜數據，求出其他公司所銷售之需評估之一 個多成份藥劑之由單位空間之馬氏距離，判斷該其他公司 之多成份藥劑與基準群有多少程度上差異之型態。另外， 選擇一定的基準群。將得自該基準群之單位空間，全國性 地或國際性一致地決定時，因爲可全國性地或國際性地一 律客觀地評估漢方製劑等之多成份藥劑，所以適宜。 本發明中，將上述之指紋圖譜數據適當地壓縮亦適宜 。亦即，將指紋圖譜數據之資訊量任意地減量亦適宜。本 發明中，以壓縮訊號強度、多成份藥劑之號碼（批號）、 溶離時間及檢測波長等之各數據（測定點）之數量爲宜。 尤其對於溶離時間，因爲有效成份由管柱所溶出的時間範 圍係有限定，所以依選擇僅該溶離時間帶之數據等而壓縮 爲宜。關於檢測波長，亦因爲有效成份之吸收光譜具有限 定的波長範圍，所以依選擇僅該限定的波長範圍之數據等 -15- (12) (12)1342393 而壓縮爲宜。另外，數據的壓縮係除了限定數據之選擇範 圍以外，亦可由許多的數據點，將適當數量的數據點，以 等間隔選出的方法等。 關於溶離時間，限定於有效成份由管柱開始溶出時間 至結束溶出時間之數據爲宜》 具體上關於溶離時間，可舉例如僅選擇有效成份由管 柱溶出之時間範圍之數據，以及通常市售之3D- HPLC裝 置雖每隔0.3至1秒之溶離時間，可得到訊號強度等數據， 但由該數據，以僅採取溶離時間之選自約每隔3秒至30秒 之等間隔秒數之數據的方法爲宜，以採取選自約每隔5秒 至2 0秒之等間隔秒數之數據的方法等尤佳。 另外，將通常所得之3D - HPLC指紋圖譜數據之數據 點，減成1/4至丨/100爲宜。另外，作爲數據點之數量係以 100至1000個爲宜，以200至800個尤佳》 關於檢測波長，以選擇重要成份具有特徵的吸收之波 長範圍爲宜。例如限定僅200nm至4 00nm之數據爲宜，以 限定200nm至300nm尤佳。 另外，將通常所得之3D- HPLC指紋圖譜數據之數據 點，減成1/2至1/50爲宜。另外，作爲數據點之數量係以 1 〇至100個爲宜，以20至80個尤佳。具體上，通常由3D-HPLC裝置所得之每隔lnm順序的波長之訊號強度等之數 據，以採取僅波長之選自每隔2nm至25ηιτι之一定波長間 隔爲宜，以採取僅選自每隔5nm至20nm之一定波長間隔 之數據尤佳之方法等。 -16- (13) (13)1342393 ±述多成份藥劑之評估方法，可適合使用於品質管理 °此時’成爲品質管理對象之一個多成份藥劑之由單位空 間之馬氏距離比選自一·定範圍的値大時，該一個多成份藥 劑爲不合格之品質管理方法尤佳。另外，上述所謂一定範 圍雖依多成份藥劑之種類而異，但可舉例以2至1 000爲宜 ，以1 0至1 00之範圍尤佳。 另外，上述多成份藥劑之評估方法，可適合使用於多 成份藥劑之製造。此時，由單位空間之馬氏距離係成爲比 選自一定範圍的値以下製造爲宜。另外，上述所謂一定範 圍，雖依多成份藥劑之種類而異，但可舉例以2至1000爲 宜，以1 0至1 0 0之範圍尤佳。 具體上可舉例如，僅於需評估之一個多成份藥劑之由 單位空間之馬氏距離爲某値以下時，選擇該一個多成份藥 劑爲製品之多成份藥劑之製造方法，或成爲某値以下之多 數個多成份藥劑或混合此等原料之多成份藥劑之製造方法 等。 本發明之評估方法係可將該步驟作爲程式爲電子計算 機所讀取而計算。 亦即，爲使電子計算機至少實施步驟（A )至（D ) (A)保存形成基準群之多數個多成份藥劑之由3D-Η P L C所得之指紋圖譜數據的步驟 (Β)輸入需評估之多成份藥劑之由3D- HPLC所得之 指紋圖譜數據，與所保存之上述（Α )之指紋圖譜數據糸且 合成一個數據群的步驟 -17- (14) (14)1342393 (c)關於上述（B )之數據群，對於多成份藥劑之號碼 與溶離時間或檢測波長，分配爲Μ T法中之變數軸’分配 訊號強度爲Μ Τ法中之特徵量，使用Μ Τ法得到單位空間 的步驟 (D)使用ΜΤ法，得到需評估之多成份藥劑之由上述 單位空間之馬氏距離的步驟 之一元化地判定一個多成份藥劑與多數個多成份藥劑 之差異程度之程式係適合使用。 ® 亦可活用爲記錄此程式之資訊記錄媒體或傳送此程式 之資訊傳送媒體。在此，所謂的資訊記錄媒體係指常用電 子計算機可讀寫之收藏資訊步驟（半導體記億體、軟碟及 硬碟等）、光讀取步驟（CD-ROM (唯讀光碟）及DVD (數位影音光碟）等），所謂資訊傳送媒體係指將程式資 訊以搬送波傳遞供給用之電腦網路（LAN (區域網絡）、 國際網路等之 WAN (廣域網路）及無線通信網路等）以 及系統中之通信媒體（光纖或無線電路等）等。 ® 另外，上述記錄媒體中，除了上述程式外，亦可記錄 基準群之數據。 . 亦即，記錄（I )及（Π ) (I )實行上述步驟（A)及步驟（B)之程式 (Π)構成基準群之多數個同種類多成份藥劑之由 3D — HP LC所得之指紋圖譜數據 之資訊記錄媒體。 另外，各多成份藥劑，例如對於各漢方製劑，僅記錄 -18- (15) (15)1342393 上述（Π )之指紋圖譜數據之記錄媒體對於評估亦有用的 〇 依據本發明，因爲不將Η P L C波峰進行波形處理，所 以數據分岐少’信賴性高’資訊量（數據點數）不局限於 特定成份之波峰數，所以可自由地增減資訊量，另外，因 爲無須組合多數成份含量之數値而判定，以—個數値即可 判定，所以可簡單地將需評估之一個多成份藥劑與基準群 之相異程度，正確地、客觀地、簡便地及一元化地判定。 【實施方式】 實施例 以下係舉例實施例及試驗例，更加詳細地說明本發明 ，但本發明並不局限於此等者。 實施例1 [步驟（1 )及步驟（2 )] 對於Tsumura社股份有限公司於1 998年至2003年期間 所製造之3 44個批號之桂枝茯苓九漢方製劑（萃取顆粒） (以下簡稱爲「TJ-25」）及A社製之桂枝茯苓九漢方 製劑，以下述方法調製測定試樣，以下述條件取得3 D — Η P L C指紋圖譜數據》 〈測定S式樣之調製方法〉 將3g之各個批號之萃取顆粒與100ml之甲醇混合，以 -19- (16) (16)1342393 句化器攪拌1 〇〇秒。放置2分鐘後，由液面與底面之中間位 置採取萃取液，以附有前過濾器之膜濾器過濾而調製測定 試樣。 〈取得3 D — HPLC指紋圖譜數據之條件〉 測定裝置：LC 一 VP系統（島津製作所社製） 分離管柱：TSK - GEL 80TS (Tosoh 社製） 移動相：A液50mM醋酸-醋酸銨緩衝溶液 φ B液乙腈1342393 (1) 玖 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 发明 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 The data measured by liquid chromatography (hereinafter referred to as "3D-HPLC") is subjected to a specific analysis step, and the evaluation method of the obtained multi-component drug is used. [Prior Art] Quantitative and qualitative evaluation of multi-ingredients, especially preparations of natural ingredients such as Kampo preparations, is based on geological factors, ecological factors, collection periods, collection sites, collection ages and growth periods of raw materials. It changes due to reasons such as the weather. Therefore, in order to evaluate the quality, safety, and efficacy of multi-component medicines such as Kampo preparations, a certain judgment standard is established, and based on this standard, national supervisory bodies, chemical organizations, and manufacturers evaluate them. However, the criterion for determining the quality of the multi-component drug or the like is generally determined by appropriately selecting one of the one or more multi-component drugs and based on the content. For example, in the case of the pharmacy νο1·28, No. 3, 67-71 (1986), when a multi-component drug cannot be used to identify an essential component, it is selected to have a quantitative analysis, is easily dissolved in water, and does not decompose in hot water. The chemical composition of the components of the chemical composition, which does not chemically react with other components, is based on the number obtained. (2) (2) 1342393 On the other hand, a chromatographic method for measuring a multi-component drug is also known, and an ultraviolet visible light absorption spectrum (hereinafter referred to as "fingerprint data") for each dissolution time is obtained, and the information of the component is used for evaluation. Benchmark. For example, it is described in Japanese Laid-Open Patent Publication No. 20 02-2, No. 2, No. 2, No. 5, a plurality of peaks are selected from the fingerprint data, and the multi-component medicine is evaluated by barcode. However, in the above method, there is a concept of "the content of a specific component" or "the chromatographic peak of a specific component", and for the quantification thereof, the chromatographic peak of the specific component is separated from the chromatographic peak of the other component on the chromatographic recording table. Or in order to find the peak area or height of a specific component, it will need to be processed by the waveform of the computer. This is one of the reasons for reducing the correctness of the data. That is, the dissolution time of the peak is not reproducible in the strict sense. Therefore, the waveform processing of the peak occurs dispersion. In particular, this separation is more pronounced when the waveform is processed with a small broad-bottom peak or a continuous peak, and there is a lack of reliability as an evaluation method. In addition, when waveform processing is performed by an electronic computer, it takes a large amount of time. Further, in the above method, since the amount of information (the amount of data points) is limited to the number of peaks of a specific component, the amount of information cannot be freely increased or decreased, and since the processing time cannot be adjusted, the most suitable evaluation method cannot be established. In addition, in the above method, since a plurality of specific component contents are obtained as a plurality of numbers, it is necessary to judge by combining these numbers, and it is not possible to evaluate a multi-component drug by one number. That is, a multi-component pharmaceutical agent to be evaluated cannot express the degree of difference from a majority of the multi-component pharmaceutical group by a number. Here, it is required to reduce the number of points, and the reliability is high, and the evaluation method of the multi-component medicine which can be judged by a number of times can be improved by processing the data time -6-(3) (3) 1342393 '. DISCLOSURE OF THE INVENTION The inventors of the present invention have tried to review the results in view of the above circumstances, and found that using the fingerprint data of 3D-HPLC, the items of the fingerprint data are processed in a specific manner, and mathematical analysis is performed to The present invention has been accomplished by a method for evaluating the quality, safety, efficacy, and the like of a multi-component pharmaceutical. That is, the present invention provides the step (2) of obtaining the fingerprint data of the 3D-HP LC of the multi-component agent to be evaluated by at least the steps (1) to (5) (1). Fingerprint data, combined with fingerprint data of 3D-HPLC of other multi-component pharmaceuticals of the same reference group (3) Regarding the fingerprint data of (2) above, for the number of the multi-component pharmaceutical, and any one The dissolution time or the detection wavelength is assigned to the variable axis in the MT method (Mahalan obis Taguchi Method), the signal intensity is the characteristic quantity in the MT method (4), and the characteristic quantity of the above (3) Step of obtaining a unit space using the ΜT method (5) Mahalanobis distance obtained by the step of obtaining the Mahalanobis distance of the multi-component medicine to be evaluated by the enthalpy method from the unit space obtained in the above (4) (4) (4) 1342393 The evaluation method of the multi-component pharmaceutical agent characterized by the degree of difference between the multi-component pharmaceutical agent to be evaluated and the majority of the multi-component pharmaceutical agent selected as the reference group. Further, the present invention provides a fingerprint of the above (1) using the step (2) of obtaining fingerprint data of 3D_HP LC of the multi-component agent to be evaluated by at least steps (1) to (8) (1). The data, the step of combining the fingerprint data of the 3D-HPLC of the same type of multi-component agent constituting the reference group (3) regarding the fingerprint data of the above (2), for the number of the multi-component drug, and any one of The dissolution time or the detection wavelength is assigned to the variable axis in the MT method (magnetic particle detection method), and the signal intensity is the characteristic quantity in the MT method (4). The characteristic quantity of the above (3) is obtained by the MT method. Step (5) of the unit space obtained by the above (4), at each detection wavelength or elution time, using the MT method to obtain the Mahalanobis distance of all multi-component medicaments (6) The other step of the step (3) is not assigned to the variable axis, and is assigned to the variable axis in the MT method, and the Mahalanobis distance obtained in the step (5) is the feature quantity in the MT method (7). The feature quantity, using the MT method to get the second order Step (8) The Mahalanobis distance obtained by the step (5) (5) 1342393 of the second unit space obtained in the above (7) using the MT method to obtain the Mahalanobis distance of the multi-component drug to be evaluated, and determining the above The method for evaluating multi-component agents characterized by the degree of difference between the multi-component agent to be evaluated and the majority of the multi-component agents selected as the reference group. Further, the present invention provides a recording medium for performing the above-described evaluation method and/or recording the reference group fingerprint data for the unit space. BEST MODE FOR CARRYING OUT THE INVENTION The present invention is a method for evaluating a multi-component medicament by using fingerprint data from a Markov distance obtained by mathematical analysis using an MT method. In the method of the present invention, a method of assigning a variable axis to the first step including the steps (1) to (5) (hereinafter referred to as "the first method of the present invention"), and performing the steps (1) to (8) are carried out 2 A method of assigning a variable axis (hereinafter referred to as "the second method of the present invention"). The first method of the present invention is carried out first as step (1), and it is necessary to obtain fingerprint data of 3D-HP LC of the multi-component drug. Here, the term "multi-component drug system" is defined as a drug containing a plurality of effective chemical components, and is not particularly limited, and is, for example, a plant extract, a crude drug, or a Chinese preparation using the same. Further, the dosage form is not particularly limited to include, for example, decoction, Hanfang extract granules, Hanfang extract, Hanfang lozenge, and Hanfang capsule. The 3D_HPLC used in the present invention refers to a spectrum of the eluted components obtained at each of a plurality of predetermined dissolution times. As a column used in this 3D-HP LC, it is usually used by users of high-performance liquid (6) (6) 1342393 phase chromatography. In addition, the detection wavelength of 3D-HP LC is not particularly limited, but is preferably in the range of 150 nm to 900 nm, and particularly preferably in the ultraviolet-visible absorption field of 200 nm to 400 nm, and is preferably selected from a plurality of wavelengths selected from 200 nm to 300 nm. . Regarding the signal intensity, transmittance or absorance of the spectrum, it is preferable to use absorbance. The fingerprint obtained in the step (1) means that the above-mentioned 3D-HPLC is used, for example, for the elution time and the detection wavelength, and the signal intensity of the above spectrum is expressed in three dimensions, for example, as shown in Fig. (a). Therefore, the so-called fingerprint data is at least a multi-component pharmaceutical number (batch number), dissolution time, detection wavelength, and signal intensity as data. Fingerprint data using 3D _HPLC can also be obtained by a commercially available device, and as a related commercially available device, for example, an LC-VP system manufactured by Shimadzu Corporation, etc., the step (2) in the first method of the present invention is as described above ( 1) The step of combining the obtained fingerprint data with the fingerprint data of 3D-HPLC of other multi-component pharmaceuticals of the same group. The fingerprint data of 3D-HPLC of other multi-component pharmaceuticals of the same type constituting the reference group can also be obtained by the above method. The difference between the measurement conditions and the like is not particularly limited, but it is preferable to use the same column for the multi-component agent to be evaluated and the same type of multi-component agent constituting the reference group under the same conditions. The other multi-component medicines of the same type constituting the reference group refer to, for example, the Chinese-10-(7) 1342393 prescription, even if they are called by the same name, the environment is different depending on the place, the collection age, and the growth period. Factors and climatological factors, etc., so even if these factors are different, most of the multi-component reference groups are selected from most multi-component pharmaceuticals, but the multi-component pharmaceutical group is suitable. Further, it is preferred to obtain all of the multi-component drug groups. In addition, when it is understood that it is a geological factor or a waiting factor for a standard, it is preferable to select a factor group as a reference group. For example, it may be limited to take the year, or may be selected by a specific company for a certain period of time. In addition, it is particularly preferable to have a certain pharmacological effect and to make it a base drug. In addition, it is preferable to select one or more of the multi-component pharmaceuticals sold in the same name, and the amount of the multi-component pharmaceuticals that constitute the reference group is not more than five, and more preferably 100 or more. The step (3) in the first method of the present invention is the step of assigning the number of the multi-component drug to the variable axis of the MT method as a signal strong characteristic amount. Here, the MT method of the present invention refers to the calculation method commonly known as the quality work, for example, "Quality Specification Association issued (2 0 00), pages 136 to 138, product collection period, collection according to geological factors The composition changes [pharmacy. The ecological factors of the fingerprint data of a certain company and the multi-component pharmaceuticals of the gas circumference. In addition, the multi-component pharmaceuticals should be suitable for most of the majority of companies. It is necessary to evaluate the benchmark group. The special limitation is that the time or the detected waviness in the fingerprint data is the industry in the MT method, and the MT method is the mathematics of the study." Japanese Quality Engineering Application Lecture-11 - (8) ( 8) 1342393 "Technology Development of Chemistry, Pharmacy" Biology, Japanese Standards Association (1999), pp. 454-456 and Quality Engineering, 丄丄(5), 78_84 (2003). In addition, a commercially available MT method software can also be used. As a commercially available MT method software, for example, PRAT for Research V1.0, PRAT for Development V].0; (Finance) 曰This specification association TM — ANOVA ; (share) Ohken MTS V e r. 2 · 0 for E xce 1 and Μ T forwindows. In the first method of the present invention, it is necessary to assign at least the number of the multi-component drug, the elution time or the detection wavelength of any one to the variable axis in the MT method, and the signal intensity to be the characteristic amount in the MT method. Although there is no particular limitation on the distribution of the variable axis, the distribution of the signal strength is assigned to the so-called number column axis (y-axis) by assigning the dissolution time to the so-called item axis (X-axis) in the ΜT method. The so-called feature quantity in the mt method is suitable. Here, the item axis (X axis) and the number column axis (y axis) are defined as follows. That is, in the MT method, 'About the data set xi in Table 1 below', the average 値mj and the standard deviation σ j ' are obtained from the normalized Χί·ΐ値 xij== (X ij - mj ) / σ j Find the correlation coefficient r ' between i and j to obtain the unit space or Mahalanobis distance. At this time, the project axis (x-axis) and the number column axis (y-axis) are as follows: "average 値mj and standard deviation<7 j series After each item axis (x-axis), the number column axis (y-axis) is changed and found. (9) 1342393 Table 1 Project axis (X axis) Number column axis (Y axis) Item 1 Item 2 ··· Item j item k Multi-component drug number 1 Multi-component drug number 2 Multi-component drug number i Multi-ingredient The number of the medicament η X ! 1 X 1 2 .·· ... X , k X2I X22 ···· X 2 k ... ····· X jj ... Xnl Xn2 ·" ·* Xnk In the first method of the present invention The step (4) is a step of obtaining a reference point and a unit amount (hereinafter simply referred to as "unit space") by using the data and the feature amount assigned to the axis in the step (3). Here, the reference point, the unit amount, and the unit space are defined in the above-described method of the method of the Μ method, and the step (5) in the method of the present invention is to obtain the multi-component medicinal agent to be evaluated by the sputum method. The step of the Markov distance of space (denoted as D2 in the method). Here, the Mahalanobis distance (D2) is defined in the same manner as the above-mentioned literature description, and the Mahalanobis distance is obtained by the method described in the above document. When the number of the multi-component drug and the dissolution time are selected as the variable axis, the Mahalanobis distance of each multi-component drug at each detection wavelength is obtained, and when the number of the multi-component drug and the detection wavelength are selected as the variable axis, each dissolution time is obtained. The Mahalanobis distance of a multi-component agent. For example, the distribution of the dissolution time on the project axis (X-axis), the number of the multi-component drug is assigned to the number column axis (y-axis) | When the signal intensity is assigned to the feature quantity in the MT method, the detection wavelength is -13 - (10) (10) 1342393 Mahalanobis distance (D2). Using the Markov distance thus obtained, it is possible to appropriately determine the degree of dissimilarity between the multi-component agent to be evaluated and the plurality of multi-component agents selected as the reference group. Further, the second method of the present invention is applied to a method generally referred to as a split synthesis method, a composite MT method, or a multi-order MT method. When the second method of the present invention is carried out, steps (1) to (5) are carried out in the same manner as in the first method of the present invention. Further, in the second method of the present invention, the step (6) is assigned to the variable axis in the MT method for the number of the multi-component drug and the other item in the step (3) which is not assigned as the variable axis, and the Mahalanobis distance. It is the step of the feature quantity in the MT method. For example, in the method (3), the solubilization time is assigned to the so-called project axis (X axis) in the ΜT method, the number of the multi-component drug is assigned to the so-called number column axis (y-axis), and the signal intensity is assigned to the MT. In the so-called feature quantity in the method, the Mahalanobis distance for the number of each multi-component drug is obtained at each detection wavelength, and secondly, the detection wavelength is assigned to the item axis (X-axis), and the number of the multi-component drug is assigned to the so-called number column axis. (y-axis) * A method of allocating the so-called feature amount in the MT method by the Mahalanobis distance obtained in the above steps (1) to (5). The step (7) in the second method of the present invention is the same as the above-described step (4), in which the second unit space is obtained based on the data group newly allocated to the axis. The step (8) in the second method of the present invention is the second unit space of -14 (11) (11) 1342393 obtained in the above (7), and the Markov distance of the multi-component agent to be evaluated is obtained by the MT method. Step: The Mahalanobis distance obtained by the step (8) (hereinafter referred to as "synthetic Markov distance") can be appropriately determined to be different from the above-mentioned one multi-component medicine to be evaluated and the plurality of multi-component medicines selected as the reference group. degree. When the synthetic Mahalanobis distance is used, it is particularly suitable because it can be evaluated by considering the total elution time, the detection wavelength, and the number of the multi-component drug, and the characteristics of 3 D - η p L C are utilized. As a suitable form of the present invention, a multi-component pharmaceutical agent manufactured or sold in advance by a specific company under the same name constitutes a reference group, and 3D-HPLC fingerprint data is obtained, and a multi-component to be evaluated by another company is obtained. The Markov distance of the pharmacy from the unit space determines how much the multi-component pharmacy of the other company differs from the reference group. In addition, select a certain reference group. When the unit space of the reference group is determined in a nationally or internationally consistent manner, it is suitable because the multi-component agent such as the Chinese preparation can be objectively evaluated nationally or internationally. In the present invention, it is also suitable to appropriately compress the above-described fingerprint data. That is, it is also appropriate to arbitrarily reduce the amount of information of the fingerprint data. In the present invention, the number of data (measurement points) such as the compressed signal intensity, the number of the multi-component drug (batch number), the elution time, and the detection wavelength is preferably used. Especially for the dissolution time, since the time range in which the active ingredient is eluted from the column is limited, it is preferable to select only the data of the dissolution time band or the like for compression. Regarding the detection wavelength, since the absorption spectrum of the active component has a defined wavelength range, it is preferable to select only the data of the limited wavelength range, such as -15-(12) (12) 1342393. Further, the compression of the data may be a method of selecting a suitable number of data points at equal intervals, in addition to a selection range of the limited data. Regarding the dissolution time, it is preferable to limit the dissolution time of the active ingredient from the beginning of the column to the end of the dissolution time. Specifically, regarding the dissolution time, for example, only data of the time range in which the active ingredient is eluted from the column is selected, and usually commercially available. Although the 3D-HPLC apparatus can obtain data such as signal intensity every 0.3 to 1 second of the elution time, the data is selected to take only the interval of the separation time from about every 3 seconds to 30 seconds. Preferably, the method of data is preferably a method of taking data selected from equal intervals of seconds every 5 seconds to 20 seconds. In addition, it is preferable to reduce the data point of the commonly obtained 3D-HPLC fingerprint data to 1/4 to 丨/100. In addition, as the number of data points, it is preferable to use 100 to 1000, and 200 to 800 is preferable. Regarding the detection wavelength, it is preferable to select a wavelength range of absorption of an important component. For example, it is preferable to define data of only 200 nm to 400 nm, and it is preferable to define 200 nm to 300 nm. In addition, it is preferred to reduce the data points of the commonly obtained 3D-HPLC fingerprint data to 1/2 to 1/50. In addition, the number of data points is preferably from 1 〇 to 100, and particularly preferably from 20 to 80. Specifically, the data of the signal intensity of the wavelength of every 1 nm sequence, which is usually obtained by a 3D-HPLC apparatus, is preferably a wavelength of a certain wavelength selected from every 2 nm to 25 ηιτι, to take only a certain wavelength interval. A method in which a certain wavelength interval of 5 nm to 20 nm is particularly preferable. -16- (13) (13) 1342393 ± The evaluation method for multi-component pharmaceuticals can be applied to quality management. At this time, the ratio of the Mahalanobis distance of the unit space to a multi-component pharmaceutical agent that is the subject of quality management is selected from one. · When the range is large, the one-component multi-agent is a substandard quality management method. Further, the above-mentioned certain range varies depending on the type of the multi-component agent, but it is preferably 2 to 1,000, and particularly preferably in the range of 10 to 100. Further, the above-described method for evaluating a multi-component pharmaceutical agent can be suitably used for the production of a multi-component pharmaceutical. In this case, it is preferable that the Mahalanobis distance in the unit space is less than or equal to a certain range. Further, the above-mentioned predetermined range varies depending on the type of the multi-component agent, but it is preferably 2 to 1000, and particularly preferably in the range of 10 to 100. Specifically, for example, when the Mahalanobis distance of the unit space of one multi-component drug to be evaluated is less than or equal to a certain value, the method of manufacturing the multi-component drug of the multi-component drug as a product may be selected, or may be a certain amount or less. A plurality of multi-component pharmaceuticals or a method for producing a multi-component pharmaceutical agent in which such raw materials are mixed. The evaluation method of the present invention can be calculated by reading the steps as a program for an electronic computer. That is, in order for the electronic computer to perform at least steps (A) to (D) (A), the step (Β) of storing the fingerprint data obtained by the 3D-Η PLC forming the plurality of multi-component pharmaceuticals of the reference group is required to be evaluated. The fingerprint data obtained by 3D-HPLC of the multi-component agent, and the saved fingerprint data of the above (Α) and the synthesis of a data group are carried out in steps -17-(14) (14) 1342393 (c) regarding the above ( The data group of B), for the number and dissolution time or detection wavelength of the multi-component drug, is assigned to the parameter axis of the ΜT method, the distribution signal intensity is the characteristic quantity in the Τ method, and the step of obtaining the unit space by using the Μ method (D) Using a sputum method, one of the steps of determining the degree of difference between a multi-component agent and a plurality of multi-component agents by one of the steps of obtaining the Mahalanobis distance of the multi-component agent to be evaluated in the unit space is suitable for use. ® can also be used as an information recording medium for recording this program or as an information delivery medium for transmitting this program. Here, the so-called information recording medium refers to the steps of collecting information that can be read and written by common electronic computers (semiconductor, floppy disk and hard disk, etc.), optical reading steps (CD-ROM (CD-ROM only) and DVD ( Digital video discs, etc.), the so-called information transmission medium refers to a computer network (LAN (regional network), WAN (wide area network) such as an international network, a wireless communication network, etc.) Communication media (fiber optics or wireless circuits, etc.) in the system. ® In addition, in the above recording medium, in addition to the above program, the data of the reference group can be recorded. That is, the records (I) and (Π) (I) are carried out by the 3D-HP LC of the majority of the same-type multi-component pharmaceuticals of the reference group by performing the above steps (A) and (B). Information recording medium for fingerprint data. In addition, each multi-component agent, for example, for each Hanfang preparation, records only -18-(15) (15) 1342393. The recording medium of the above (Π) fingerprint data is also useful for evaluation, according to the present invention, because it is not The waveform of the PLC peak is processed, so the data is less distributed. The 'reliability is high' information volume (number of data points) is not limited to the number of peaks of a specific component, so the amount of information can be freely increased or decreased, and because it is not necessary to combine the content of most components. It is judged by 値, and it can be determined by a number of ,, so that the degree of dissimilarity between a multi-component drug to be evaluated and the reference group can be easily determined accurately, objectively, simply, and in a unified manner. [Embodiment] Hereinafter, the present invention will be described in more detail by way of examples and test examples, but the invention is not limited thereto. Example 1 [Step (1) and Step (2)] For the 4,44 batches of cassia twigs and kiln preparations (extracted granules) manufactured by Tsumura Co., Ltd. from 1998 to 2003 (hereinafter referred to as "TJ-25") and the Guizhi Jiuhan recipe prepared by A, the measurement sample was prepared by the following method, and 3 D - Η PLC fingerprint data was obtained under the following conditions: <Measurement method for measuring S pattern> 3g The extracted granules of each batch were mixed with 100 ml of methanol and stirred for 1 〇〇 second with a -19-(16) (16) 1342393 narrator. After standing for 2 minutes, an extract was taken from the middle of the liquid surface and the bottom surface, and filtered with a membrane filter attached with a pre-filter to prepare a measurement sample. <Acquired 3 D - Conditions for HPLC Fingerprint Data> Measurement device: LC-VP system (manufactured by Shimadzu Corporation) Separation column: TSK - GEL 80TS (manufactured by Tosoh Corporation) Mobile phase: A solution 50 mM acetic acid-ammonium acetate buffer solution φ B liquid acetonitrile
梯度條件：線性梯度 管柱溫度：4(TC 流速：l.〇mL/min 注入量：3 0以L 檢測器：SPD — MIOAvp (島津製作所社製） 測定波長：200niri至400nm 分析軟體：CLASS-VP (島津製作所社製） Φ 關於所取得3D - HPLC指紋圖譜數據’如表2所示， 進行數據壓縮。 · -20- (17) 1342393 表2 取得數據 壓縮數據方法 溶離時間 範圍 〇-80min 8 〜3 0 m i η 溶離時間 數據點 每0.6 4秒 每1 2.8秒 測定波長 範圍 2〇0~400nm 200~300nm 測定波長 數據點 每1 nm 每 1 0 n m 訊號強度 有效數字 1 pV(nAbs) 以下捨去 1 mV (m Abs) 以下捨去 依此，數據點數量如以下所示。 多成份藥劑之號碼· · · ·344個 檢測波長........1 1個 溶離時間........1 0 8個 數據壓縮前後之指紋圖譜之1例，如圖1所示。由此而 無問題地可以ΜΤ法計算單位空間及馬氏距離。 [步驟（3 )] 關於所取得之3 D - HPLC指紋圖譜數據，如表3之右 欄所示，進行分配各檢測波長於ΜΤ法中之軸。 作爲「各檢測波長」係將200nm至3 00nm之範圍，每 l〇nm等分割所得之各1 1個波長。 -21 - (18) 1342393 表3 取得指紋圖譜數據 軸之分配後 軸之 分配前 於每多成份藥劑之號碼 於每檢出波長 項目軸（X軸） 檢測波長 項目軸（X軸） 溶離時間 號碼列軸（y軸） 溶離時間 號碼列軸（y軸） 多成份藥劑之號碼 特徵量 訊號強度 特徵量 訊號強度 依此，通常所取得之指紋圖譜數據成爲適合MT法之 數値分析之形式。 [步驟（4 )] 如「品質工學的數理j曰本規格協會發行（2000 )第 1 3 6至1 3 8頁，品質工學應用講座「化學、藥學、生物學之 技術開發」日本規格協會編（1 999 )第45 4至456頁及品質 工學，1丄（5) ，7 8 - 84 ( 2003年）所記載之使用mt法 ，而得單位空間亦即基準點與單位量。 I十算係使用（股）Ohken社之MT f〇rwindows〇 [步驟（5 )] 將A社製之桂枝获令九漢方製劑之由步驟（4 )所得； 之基準點之馬氏距離，依據上述文獻記載，依MT法# $ 出。 其次，步驟（1 )中，除了取代成A社製之桂枝侠爷 -22- (19) (19)1342393 九漢方製劑，表4中所記載之其他1 2個爲「需評估之一個 多成份藥劑」之外，與上述步驟（1 )至步驟（5 )同樣地 計算各桂枝茯苓九漢方製劑之馬氏距離。 馬氏距離係將200nm至3 00nm之範圍，每i〇nm等分 割所得之各Π個波長’於2 3 Onm之馬氏距離一例如表4所 示。 表4 桂枝茯苓九 馬氏距離 TJ-25-I(Tsumura 社製） 14.9 TJ-25-II(Tsumura 社製） 13.1 A社製 22 7.0 B社製 64.0 C社製-1 407.7 C社製-11 6 5 3.2 C社製· III 8 1.3 C社製-1V 113.8 C社製-V 96.7 D社製 33.7 E社製 28.0 F社製 84.0 G社製 1 476.4 依據使用以步驟（1 )至步驟（5 )所得之各檢測波長 之馬氏距離’可判定某一個桂枝茯苓九與基準群之相異程 -23- (20) 1342393 度。由此顯示此評估方法對於品質管理係有效的。 實施例2 實施例1中，於各檢測波長，可得到對於各多成份藥 劑之號碼之馬氏距離，由實施例1所得之馬氏距離作爲MT 法中之特徵量，由如下所示之多階MT法求出合成馬氏距 離。 [步驟（】）及步驟（5 )] 與實施例1同樣地進行。 [步驟（6 )] 如表5所示，進行分配MT法中之軸。 表5 軸之分配 項目（X軸） 檢測波長 號碼列軸（y軸） 多成份藥劑之號碼 特徵量 實施例1所得之馬氏距離Gradient conditions: Linear gradient column temperature: 4 (TC flow rate: l. 〇mL/min Injection volume: 3 0 to L Detector: SPD - MIOAvp (made by Shimadzu Corporation) Measurement wavelength: 200niri to 400nm Analysis software: CLASS- VP (made by Shimadzu Corporation) Φ About the obtained 3D-HPLC fingerprint data 'As shown in Table 2, data compression is performed. · -20- (17) 1342393 Table 2 Data compression data method solvation time range 〇-80min 8 ~3 0 mi η Dissolution time data point every 0.6 4 seconds every 1 2.8 seconds Measurement wavelength range 2〇0~400nm 200~300nm Measurement wavelength data point per 1 nm every 10 nm Signal strength Effective number 1 pV(nAbs) To go below 1 mV (m Abs), the number of data points is as follows. Multi-component drug number · · · · 344 detection wavelengths........1 1 dissolution time... .....1 0 1 image of the fingerprint before and after data compression, as shown in Figure 1. Therefore, the unit space and the Mahalanobis distance can be calculated without any problem. [Step (3)] The obtained 3 D - HPLC fingerprint data, as shown in the right column of Table 3, was divided into Each of the detection wavelengths is in the axis of the enthalpy method. As the "detection wavelength", each of the wavelengths is divided into the range of 200 nm to 300 nm, and each of the wavelengths is divided into 1 to 1 nm. -21 - (18) 1342393 Table 3 The distribution of the map data axis is assigned to the axis of each multi-component agent before each detected wavelength item axis (X-axis). Detection wavelength Item axis (X-axis) Dissolution time number column axis (y-axis) Dissolution time number column axis (y-axis) Multi-component pharmacy number characteristic quantity signal intensity characteristic quantity signal strength According to this, the commonly obtained fingerprint map data becomes a form suitable for the number analysis of the MT method. [Step (4)] Such as "Quality Engineering"数 曰 曰 规格 规格 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( And quality engineering, 1 丄 (5), 7 8 - 84 (2003) using the mt method, and the unit space is also the reference point and unit amount. I ten system use (shares) Ohken Society MT F〇rwindows〇[Step (5)] Guizhi was obtained from step (4) of Jiuhanfang preparation; the Mahalanobis distance of the reference point was recorded according to the above document according to the MT method. Secondly, in step (1), except for the substitution of A system.桂枝侠爷-22- (19) (19) 1342393 Jiuhan prescription preparation, the other 12 items listed in Table 4 are "one multi-component medicine to be evaluated", and the above steps (1) to ( 5) Calculate the Mahalanobis distance of each cassia twig and Jiuhan prescription in the same manner. The Mahalanobis distance is in the range of 200 nm to 300 nm, and each of the wavelengths obtained by dividing each i 〇 nm is a Mahalanobis distance of 2 3 Onm as shown in Table 4, for example. Table 4 桂 茯苓 茯苓 马 马 马 马 马 马 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 14. 11 6 5 3.2 C system system III III 1.3 C company system -1V 113.8 C company system -V 96.7 D company system 33.7 E system system 28.0 F system system 84.0 G system system 1 476.4 Use step (1) to step ( 5) The Markov distance of each detection wavelength obtained can determine the phase difference -23-(20) 1342393 degrees between a certain cassia twig and the reference group. This shows that this evaluation method is effective for quality management. [Embodiment 2] In Example 1, the Mahalanobis distance of the number of each multi-component drug was obtained at each detection wavelength, and the Mahalanobis distance obtained in Example 1 was used as the feature amount in the MT method, as shown below. The synthetic Markov distance is obtained by the order MT method. [Step ( )) and Step (5 )] The same procedure as in Example 1 was carried out. [Step (6)] As shown in Table 5, the axis in the MT method is assigned. Table 5 Axis assignment Item (X-axis) Detection wavelength Number column axis (y-axis) Multi-component drug number Characteristic amount The Mahalanobis distance obtained in Example 1
亦即，如表6所示之數據組。 -24- (21) 1342393 表6 號碼軸（y軸） 多成份藥劑之號碼 項目軸（X軸）檢測波長 2 0 0 n m 2 1 0 n m • > · i • · « 3 0 0 n m 1 (D 1 . 2 0 0 )2 (D 1.21 〇)2 • * · (D.-i)2 ♦ · · (D 丨.3 0 0 )2 2 (D 2 . 2 0 0 ) 2 (D2.210)2 > · · (D2.i)2 •, · (D2.300)2 • · » * · * ♦ · · ♦ · * • · · ,· · • · · i (〇|.20〇)2 (Dj.2 丨 0)2 … (Di.i)2 ... (Di.30〇)2 * · * • * · ..· * · ... .,. ..· 344 (D344.200)2 (D344.2IO)2 … (D344 i)2 … (D344· 300)2That is, the data set as shown in Table 6. -24- (21) 1342393 Table 6 Number axis (y-axis) Multi-component drug number item axis (X-axis) detection wavelength 2 0 0 nm 2 1 0 nm • > · i • · « 3 0 0 nm 1 ( D 1 . 2 0 0 ) 2 (D 1.21 〇) 2 • * · (D.-i) 2 ♦ · · (D 丨.3 0 0 ) 2 2 (D 2 . 2 0 0 ) 2 (D2.210 ) 2 > · · (D2.i)2 •, · (D2.300)2 • · » * · * ♦ · · ♦ · * · · · , · · · · · i (〇|.20〇) 2 (Dj.2 丨0)2 ... (Di.i)2 ... (Di.30〇)2 * · * • * · ..· * · ... .,. ..· 344 (D344. 200) 2 (D344.2IO) 2 ... (D344 i) 2 ... (D344· 300) 2
表中，（Di，j ) 2係多成份藥劑之號碼i，檢測波長j 時之實施例1所得之馬氏距離。 [步驟（7)及步驟（8)] 與實施例1同樣地求出單位空間，算出合成馬氏距離 〇 圖2表示，同樣地爲得到單位空間，以所使用之 Tsumura社製桂枝夜茶九（TJ—25)之344個批號之合成 馬氏距離爲橫軸，其具有馬氏距離之試樣個數爲縱軸所作 成的圖。另外，爲得到單位空間所使用之Ή - 25之344個 批號之馬氏距離之平均値，依其定義經常爲1。 符合需評估之一個多成份藥劑之2個Tsumura社製桂 枝茯苓九與Tsumura社製以外之1 1個桂枝茯苓九漢方製劑 之由基準點之合成馬氏距離如表7所示。 -25- (22) 1342393 表 7 桂枝获茶九 合成馬氏距離 TJ-25-I(Tsumura 社製） 63.3 TJ-25-II(Tsumura 社製） 222.2 A社製 2397.6 B社製 1110.0 C社製-I 4911.5 c社製-II 7496.4 c社製-III 3226.2 c社製-IV 5546.2 c社製_v 3489.5 D社製 2 3 8.6 E社製 462.8 F社製 890.7 G社製 50698.5In the table, (Di, j ) 2 is the number i of the multi-component drug, and the Mahalanobis distance obtained in Example 1 when the wavelength j is detected. [Step (7) and Step (8)] The unit space is obtained in the same manner as in the first embodiment, and the synthetic Mahalanobis distance is calculated. FIG. 2 shows that the unit space is obtained in the same manner, and the Tatsuura tea made by Tsumura Co., Ltd. is used. The synthetic Markov distance of the 344 batch numbers of nine (TJ-25) is the horizontal axis, and the number of samples having the Mahalanobis distance is the vertical axis. In addition, the average 马 of the Markov distances of 344 batch numbers used to obtain the unit space is often 1 according to its definition. The synthetic Mahalanobis distance from the reference point of the two Tsinghua-based Kyosuke Nippon and the Tsumura Co., Ltd., which are one multi-component drug to be evaluated, are shown in Table 7. -25- (22) 1342393 Table 7 Cassia Cuiji Jiuji Synthetic Markov Distance TJ-25-I (Tsumura Co., Ltd.) 63.3 TJ-25-II (Tsumura) 222.2 A company 2397.6 B company 1110.0 C System-I 4911.5 c-system-II 7496.4 c-system-III 3226.2 c-system-IV 5546.2 c system _v 3489.5 D company system 2 3 8.6 E system system 462.8 F system system 890.7 G system system 50698.5
由使用所得之合成馬氏距離，與344個Tsumura社製 桂枝获茶九漢方製劑（TJ — 25 )之相異程度係可以一個數 値判定。其次，亦顯示例如G社製或C社製-Π係與3 4 4 個之TJ- 2 5所成之基準群差異最大等。 另外’使用合成馬氏距離，可由廣泛地檢測波長範圍 而全面性地評估。 進而’使用此評估方法，明白可製造品質安定之桂枝 茯苓九漢方製劑。 -26- 1342393 第093116466號專利申請案中文說明書修正頁 修正替换寅 民國100年2月18日修正 另外，關於此桂枝茯苓九漢方製劑（TJ - 25 ) ，MT法 之單位空間之數據所記錄之記錄媒體對於評估係有用的。 實施例3 對於Tsumura社股份有限公司於1 998年至2003年期間 所製造之100個批號之補中益氣湯（以下簡稱爲「TJ - 41 J )，以與實施例1及2相同的方法調製測定試樣，以相同 的條件取得3D — HPLC指紋圖譜數據，使用MT法而得到 單位空間，亦即基準點與單位量。 爲得到上述單位空間所使用之全部TJ - 4 1與6個試樣 之TSUmura社製以外的補中益氣湯之由單位空間之馬氏距 離。使用馬氏距離時，明白可簡單地、正確地及客觀地判 斷一個補中益氣湯之由基準點之差異程度。 圖3 (a)係表示馬氏距離爲1.0之補中益氣湯（TJ-41)之指紋圖譜。另外，圖3(b)係表示馬氏距離爲2704 之補中益氣湯之指紋。若只見指紋，此等之相異程度並不 清楚，但由本發明所得之馬氏距離，從如1.0與2704數値 差異之大，明白馬氏距離對於定量地判斷多成份藥劑彼此 間的差異用係有效的。 產業上利用性 依據本發明，因爲信賴性高，可自由地增減資訊量， 所以可適當地縮短處理數據的時間，可以一個數値簡單地 判定需評估之一個多成份藥劑與基準群之差異。因此，因 -27- 1342393 爲容易地評估多成份藥劑之相異程度， 之品質管理者。另外’使用其而可適當 〇 另外，依據預先以本發明之單位空 際性之統一基準，係可判定某多成份藥 者，可供給安定品質之漢方製劑等之多 【圖式簡單說明】 圖1係表示數據壓縮前後之桂枝茯 圖譜圖。圖中係表示（a)爲數據壓縮補 縮後者。 圖2係表示344個桂枝茯苓九漢方製 TJ - 25 )之馬氏距離與個數之關係圖》 圖3係表示馬氏距離爲1〇之補中益 製TJ—41)與馬氏距離爲2704之補中 圖。圖中係表示（a)馬氏距離爲1.0, 2 7 04 者。 所以可使用於社內 地製造多成份藥劑 間定義全國性或國 劑與該基準之相異 成份藥劑。 苓九漢方製劑指紋 ίϋ，（ b )爲數據壓 劑（T s u m u r a社製 氣湯（Tsumura 社 益氣湯之指紋圖譜 (b )馬氏距離爲The degree of dissimilarity between the synthetic Markov distance obtained by using and the 344 Tsumura Co., Ltd., which is made by Tsinghua Co., Ltd. (TJ-25) can be determined by one number. Next, it is also shown that, for example, the difference between the reference group formed by the G system or the C system-Π system and the 344 TJ-25 is the largest. In addition, the use of synthetic Markov distance can be comprehensively evaluated by extensively detecting the wavelength range. Furthermore, using this evaluation method, it is understood that the Guizhi, Jiuhanfang preparation can be manufactured with a stable quality. -26- 1342393 Patent Application No. 093116466 Revision of the Chinese version of the amendments to the amendments of the Republic of China on February 18, 100. In addition, the data on the unit space of the MT method is recorded on this Guizhi Jiuhanfang preparation (TJ-25). The recording medium is useful for the evaluation system. Example 3 The Buzhong Yiqi Decoction (hereinafter referred to as "TJ-41 J") of 100 batches manufactured by Tsuruma Co., Ltd. during the period from 998 to 2003 was the same as in Examples 1 and 2. The measurement sample was prepared, and 3D-HPLC fingerprint data was obtained under the same conditions, and the unit space, that is, the reference point and the unit amount, was obtained by the MT method. All the TJ-4 1 and 6 tests used to obtain the above unit space were obtained. The Markov distance of the unit space of Buzhong Yiqi Decoction outside the TSUmura system. When using the Mahalanobis distance, it is clear that the difference between the benchmark points of a Buzhong Yiqi Decoction can be judged simply, correctly and objectively. Figure 3 (a) shows the fingerprint of Buzhong Yiqi Decoction (TJ-41) with a Markov distance of 1.0. In addition, Figure 3(b) shows the Buzhong Yiqi Decoction with a Markov distance of 2704. Fingerprint. If only the fingerprint is seen, the degree of difference is not clear, but the Mahalanobis distance obtained by the present invention is different from the difference between the numbers of 1.0 and 2704, and the Mahalanobis distance is used to quantitatively judge the multi-component medicaments between each other. The difference is effective. Industrial use According to the present invention, since the amount of information can be freely increased or decreased because of high reliability, the time for processing the data can be appropriately shortened, and the difference between a multi-component medicine to be evaluated and the reference group can be easily determined in a single number. -27- 1342393 is a quality manager who can easily evaluate the degree of dissimilarity of multi-component pharmaceuticals. In addition, it can be appropriately used, and it can be judged based on the unified standard of the unit space of the present invention. Ingredients, can supply more stable Chinese herbal preparations, etc. [Simplified illustration of the schema] Figure 1 shows the map of osmanthus fragrans before and after data compression. The figure shows that (a) is the data compression and the latter. The system shows the relationship between the Markov distance and the number of 344 Guizhi, Jiuhanfang TJ - 25). Figure 3 shows that the Mohs distance is 1〇, the Buzhongyi system TJ—41) and the Mahalanobis distance is 2704. In the figure, the figure shows that (a) the Mahalanobis distance is 1.0, 2 7 04. Therefore, it can be used for the manufacture of multi-component pharmaceuticals in the community to define national or national agents and different ingredients of the standard. Jiuhan Party System Fingerprint ίϋ, (b) a data-pressure agent (T s u m u r a decoction Corporation (Tsumura Co. Decoction of the fingerprint (b) is a Mahalanobis distance