目的：為了找到更有效的藥物代謝物鑑定方法，本實驗結合了時間序列實驗與穩定同位素追蹤法的代謝體學資料處理方法，以鑑定藥物代謝物。

 

方法：使用曲格列酮(troglitazone) 和其同位素藥物 (D4-troglitazone) 與人類肝酵素進行培養，產生藥物代謝物。將藥物分別培養四個不同時間：0、2、4、6 小時。利用高解析液相層析質譜儀對樣本進行分析。將所得到的質譜檔案轉換為訊號資訊。我們使用時間序列實驗結合穩定同位素追蹤法篩選可能為藥物代謝物的訊號。接著，利用劑量效應實驗，驗證這些篩選到的訊後是否呈現劑量效應。最後，使用串聯式質譜儀對代謝物訊號進行二次碰撞，並推測其可能的化學結構。結果：藥物與人類肝酵素培養樣本的質譜檔案經

軟體轉換為訊號資訊，共獲得7360 個訊號。經過第一階段篩選，有292 個訊號的校正後的面積與培養時間呈現正相關。經過第二階段篩選，這292 個訊號中有16 個是具有同位素配對的訊號。在這16 個訊號中，其中有13 個呈現劑量效應。而這13 個訊號有較高機率為TRO (troglitazone) 代謝物。通過推測這13 個訊號的結構，發現其中有2 個能夠推測出可能的化學結構，並由這些結構可驗證為TRO 的代謝物，且過去尚未報導過。

 

結論：本研究所使用的時間序列實驗結合同位素追蹤法，能夠有效地鑑定藥物代謝物。我們找到的代謝物訊號中，其中2 個是可以推測出化學結構，且未曾被報告過。此外，我們還發現了這2 個代謝物的化學結構的特徵是噻唑烷二酮(thiazolidinedione) 裂解產物，可能與肝臟毒性有關。

 

 

Objective: To develop more effective methods for identifying drug metabolites, this study combined time-course experiments with stable isotope tracing (SIT) to discover drug metabolites. Method: Troglitazone (TRO) and its isotopically labelled compound (D4-TRO) were incubated with human liver enzymes to generate TRO metabolites. The incubation was performed at four different time points: 0, 2, 4, 6 hours. The samples were analyzed using high-resolution liquid chromatography mass spectrometry, and the resulting mass spectrometry data were converted into signal lists. We employed a two-stage data processing method to filter the potential metabolite signals. A doseresponse experiment was conducted to validate whether these identified signals exhibited a dose-response. Subsequently, tandem mass spectrometry was used to analyze the identified signals and infer their possible chemical structures. 

 

Results: The mass spectrometry data from the human liver enzyme incubation samples yielded a total of 7,360 signals. In the first stage of screening, 292 signals showed a positive correlation between adjusted signal abundances and incubation time. In the second stage of screening, 16 out of these 292 signals were found to have isotopic pairs, and among these, 13 showed a dose-response relationship. These 13 signals are likely to be TRO metabolites with high possibility. Among these, the chemical structures of 2 signals were inferred, indicating that they could be previously unreported TRO metabolites. 

 

Conclusion: The combination of time-course experiments with SIT prove effective in identifying TRO metabolites. Among the metabolite signals detected, 2 of them were inferred to have chemical structures that had not been previously reported. Additionally, we discovered that the chemical structure of these 2 metabolites were cleavage products of thiazolidinedione, which may be associated with liver toxicity.

 

 

Submitted for publication: 2023.11.17; Accepted for publication: 2024.5.31