糞便潛血檢驗偽陰性之大腸直腸腫瘤-大腸直腸癌篩檢之夢魘

Abstract

背景 大腸直腸癌是已開發與開發中國家重要的公共衛生議題。同樣的，台灣也面臨著大腸直腸癌的嚴峻考驗，目前大腸直腸癌是台灣發生人數最多的癌症，同時也是癌症死亡人數的第三位，每年約有4,000人死於大腸直腸癌，每年新發生的案例也約有16,000例。因此，如何預防大腸直腸癌的發生，以及如何改善其治療效果，是我們需要面對與解決的公共衛生挑戰。大腸直腸篩檢已經被證實可以有效地同時降低大腸直腸癌的發生率與死亡率，美國從1970年代開始實行篩檢的工作，因為起頭得早，目前美國成為大腸直腸癌發生率與死亡率皆下降的國家。台灣從2007年開始實行全國性的篩檢，篩檢方式使用免疫法糞便潛血檢驗，篩檢對象是50-75歲的族群，篩檢頻率為每兩年一次。糞便潛血呈現陽性的民眾，需要緊接著接受大腸鏡當作為確診工具，然後依據大腸鏡的檢查結果，決定其治療的方式與追蹤的頻率。例如大腸鏡發現一顆一公分的息肉，可以接受大腸鏡息肉切除術移除息肉，然後三年後再接受一次大腸鏡的複察。至於糞便潛血檢查呈現陰性的患者，則兩年後再接受免疫法糞便潛血檢查即可。目前已經有研究報告顯示，採用此方式接受大腸直腸癌篩檢的民眾，和沒有接受篩檢的民眾比較起來，可以有效地降低十個百分點的死亡率。由此可見，免疫法糞便潛血檢查可以很有效地篩檢大腸直腸癌，然而，免疫法糞便潛血檢查仍然不是一個完美的篩檢工具，其對於大腸直腸癌的敏感度為79%，對於癌前病變-進階型腺瘤(advanced adenoma)的敏感度則只有29%，一旦腸內有腫瘤，但是糞便潛血檢查卻呈現陰性，便容易在篩檢後出現大腸直腸癌，也就是產生間隔癌(interval cancer)。如何降低糞便篩檢後產生間隔癌的機會，其重要性不言可諭，欲減少間隔癌的發生，很重要的工作之一，就是瞭解哪些大腸直腸腫瘤不容易被免疫法糞便潛血檢查檢出，這種腫瘤稱之為免疫法糞便潛血檢查偽陰性的腫瘤。過去的研究顯示，非隆起型的大腸直腸腫瘤容易成為糞便潛血偽陰性的腫瘤，其中，扁平鋸齒狀腺瘤(sessile serrated adenoma/polyp, SSA/P)與凹陷型大腸直腸腫瘤(depressed colorectal neoplasm)這兩種重要的大腸直腸癌前病變都屬於非隆起型的腫瘤，因此糞便潛血就不容易偵測到它們，所以這兩種癌前病變都是容易成為糞便潛血偽陰性的腫瘤，相對地，也容易造成間隔癌。 目的 此研究主要是以扁平鋸齒狀腺瘤與凹陷型大腸直腸腫瘤為主題，藉由探討其流行病學、分子變化、內視鏡特徵、與糞便潛血檢查的敏感度等。希望能夠瞭解其容易造成糞便潛血檢查偽陰性的原因，並藉此找出解決方式，提高其篩檢的診斷率，減少間隔癌發生的機會。 方法 針對扁平鋸齒狀腺瘤部分，我們利用至台大醫院健康管理中心接受篩檢性大腸鏡的民眾，挑出50至75歲並符合大腸直腸癌一般風險(average risk)的族群，分析以10、15、20 µg Hb/g feces為陽性值標準時，免疫法糞便潛血檢查對於傳統腺瘤以及扁平鋸齒狀腺瘤的敏感度，接著比較糞便潛血檢查對近端與遠端鋸齒狀腺瘤的敏感度。除了探討糞便潛血檢查對於扁平鋸齒狀腺瘤的敏感度外，我們也觀察這類腫瘤的內視鏡特徵，首先，依據文獻報告整理出六個扁平鋸齒狀腺瘤的代表性內視鏡特徵，這六個特徵包含了白光、窄頻影像系統、與染色內視鏡的特有表現，然後將所有個案按照收案順序分成測試族群(derivation cohort)與驗證族群(validation cohort) ，在測試族群中找出最佳的內視鏡特徵組合，最後在驗證族群驗證此特徵組合診斷鋸齒狀腺瘤的準確性。為了瞭解凹陷型腫瘤的基因變化，先收集凹陷、扁平、與隆起型腫瘤，然後利用PCR (polymerase chain reaction)測試KRAS、BRAF、PIK3CA在不同型態腫瘤的表現。接著，利用次世代定序來找凹陷型腫瘤的腫瘤標記，實驗組和對照組分別為凹陷型與隆起型腫瘤，首先利用OncoScan FFPE array在測試族群中找出凹陷型腫瘤的生物標記，接著在驗證族群驗證生物標記的準確度，最後，利用大腸直腸癌的組織，探討凹陷型腫瘤發展而成的大腸直腸癌，其存活表現與傳統腸癌有無不同。 結果 以10、15、20 µg Hb/g feces為陽性值標準時，免疫法糞便潛血檢查對扁平鋸齒狀腺瘤的敏感度為12.3%、6.2%、6.2%，對大型扁平鋸齒狀腺瘤的敏感度為18.4%、10.5%、10.5%，對進階型腺瘤的敏感度為32.4%、24.5%、20.9%。回歸分析顯示有扁平鋸齒狀腺瘤的民眾和無息肉的民眾比較起來，其糞便檢查陽性的風險，並不會增加。而大型扁平鋸齒狀腺瘤與進階型腺瘤比較，在標準值為15 µg Hb/g時，大型扁平鋸齒狀腺瘤的糞便潛血陽性風險顯著降低 (OR=0.3, 95%信賴區間=0.10-0.90)。因此，若欲提高扁平鋸齒狀腺瘤的診斷度，就必須先增加其內視鏡診斷率，使用染色內視鏡與窄頻影像系統可以有效改善其診斷率，其中，大型鋸齒狀腺瘤用染色內視鏡加上窄頻影像系統效果最好，如果是小型鋸齒狀腺瘤，利用染色內視鏡效果就很好。凹陷型腫瘤的基因變化和傳統大腸直腸癌不相同，傳統大腸直腸癌常出現的KRAS、BRAF、PIK3CA突變，在凹陷型腫瘤出現的機會很低。利用次世代定序發現凹陷型腫瘤特異性的生物標記組合為MYC, CCNA1, BIRC7，此生物標記組合在測試族群與驗證族群的敏感度/特異性/準確度分別為75.7/83.3/79.7%。利用此凹陷型腫瘤標記在530個大腸直腸癌個案做驗證，發現22.1%的腸癌由凹陷型腫瘤形成，凹陷型大腸直腸癌的整體progression-free survival(PFS)較非凹陷型腸癌差，如果進一步分析，凹陷型腸癌在stageＩ、T1+2, 近端大腸的PFS皆較差。 結論 免疫法糞便潛血檢查是目前第一線的腸癌篩檢工具，然而扁平鋸齒狀腺瘤和凹陷型腫瘤都是糞便潛血檢查容易偽陰性的病兆，提高這兩類病兆的診斷率，就有機會降低篩檢後的間隔癌。本研究顯示，免疫法糞便潛血檢查對扁平鋸齒狀腺瘤的敏感度不佳，即便是大型鋸齒狀腺瘤，其敏感度還是較大型傳統腺瘤明顯降低，欲提升其診斷率，應加強內視鏡的診斷率，使用染色內視鏡與窄頻影像系統可以有效提升鋸齒狀腺瘤的內視鏡診斷率。凹陷型腫瘤的基因變化目前尚未清楚，本研究證明傳統大腸直腸癌會出現的KRAS、BRAF、PIK3CA突變，在凹陷型腫瘤的癌症形成早期鮮少出現。MYC、CCNA1、BICR7是對凹陷型腫瘤具有專一性的生物標記組合，藉由此生物標記組合，我們了解由凹陷型腫瘤發展而成之腸癌的臨床特性，此外，此生物標記也提供了一個發展診斷凹陷型腫瘤工具的機會，為早期診斷凹陷型腫瘤奠下基礎。糞便潛血偽陰性的腫瘤是癌症篩檢的重要挑戰，應該藉由提高內視鏡的診斷率以及發展分子診斷工具，來分別提高扁平鋸齒狀與凹陷型腫瘤的診斷率，減少糞便潛血偽陰性腫瘤的威脅。

Background Colorectal cancer (CRC) is a challenge for public health worldwide. The incidence and mortality increase not only in western countries but also in eastern states. Nowadays, CRC is the most incident malignancy and the 3rd most common cause of cancer-related death in Taiwan. There was 16,000 newly developed CRC, and 4,000 people die of it annually in our country. CRC screening was proven effective to prevent CRC and reduce its mortality. In the US, the nationwide screening program was launched since the 1970s, and the incidence of CRC had started to decrease since 2000s. In Taiwan, the nationwide screening program was initiated since 2007. Biennial fecal immunochemical test (FIT) is offered for subjects aging 50 to 75 years old. The subject has a positive result of FIT will receive confirmatory colonoscopy subsequently, and the interval of surveillance colonoscopy will be recommended based on the colonoscopic findings. For those who have a negative result of FIT, FIT will be repeated two years later. In comparison with subjects refused to screen, the subjects participating in the FIT-based screening program could have a reduction of CRC mortality by 10%. Thus, there is no doubt that FIT could reduce the mortality effectively. However, FIT has its limitations, and one of them is limited sensitivity for detecting pre-cancerous lesion and early cancer. The sensitivity of FIT for detecting CRC and advanced adenoma is 79% and 29% respectively. Once a crucial precancerous lesion is failed to be identified by FIT, interval cancer will develop. The neoplasm which was unable to be detected by FIT is called FIT false-negative neoplasm. To understand the FIT false-negative neoplasm is a basis to reduce the post-FIT interval cancer. Both sessile serrated adenoma/polyp (SSA/P) and depressed neoplasm are the FIT false-negative neoplasm based on their anatomic location, tumor size, and morphology. Thus, we aim to investigate the diagnostic performance of FIT for detecting SSA/P, to test the usefulness of image enhancing endoscopy (IEE) for increasing diagnostic yield of SSA/P, to clarify the molecular change of depressed neoplasm and to develop the biomarker specific to the depressed neoplasm. Moreover, we also aimed to investigate the survival outcome of the de novo CRC which arise from the depressed neoplasm. Methods We enrolled the average-risk subjects aging 50 to 75 years old. These people received concurrent screening colonoscopy and FIT in the National Taiwan University Hospital. The sensitivity of FIT was investigated for detecting SSA/P, large SSA/P and advanced adenoma (AA) at cutoffs of 10、15、20 µg Hb/g feces respectively. Also, the association between positivity of FIT and each category of colorectal neoplasm was also investigated by regression analysis. For clarifying the usefulness of IEE for diagnosing SSA/P, we studied the serrated polyps with white-light endoscopy (WLE), chromoendoscopy, and narrow band imaging (NBI). The SSA/P specific endoscopic characteristics using each IEE modality alone or in combination were applied for identifying SSA/P. Regression analysis in the derivation cohort explored the best combination of the endoscopic feature in each IEE scenario. Subsequently, the best combination generated in the derivation cohort was checked for the diagnosing performance by receiver operating characteristic curve in the validation cohort. For exploring the molecular change of depressed neoplasm, we investigated the mutation of KRAS, BRAF, and PIK3CA in first and then identified its specific biomarkers through the next-generation sequence. Finally, the biomarkers were applied for clarifying the clinical impact of CRC arising from depressed neoplasm, so called de novo CRC, including the prevalence in each cancer stage and the survival outcome. Results The sensitivity of FIT for detecting SSA/P was 12.3%, 6.2% and 6.2% at a cutoff of 10, 15, and 20 µg Hb/g feces respectively. The sensitivity for detecting large SSA/P and advanced adenoma (AA) was 12.3%, 10.5%, and 10.5% and 32.4%, 24.5% and 20.9% at different cutoffs. The regression analysis demonstrated that the positivity of FIT was without difference between normal mucosa and SSA/P. Even large SSA/P, its positivity of FIT was significantly lower than AA at a cutoff of 15 µg Hb/g. (odds ratio=0.3, 95%CI=0.10-0.90) For improving the endoscopic diagnosis, the use of chromoendoscopy or NBI could increase the area under the curve in comparison to WLE alone. For small SSA/P, chromoendoscopy was the best modality for identifying SSA/P. For large SSA/P, combination use of chromoendoscopy and NBI was better than use anyone alone. Regarding the genetic change of depressed neoplasm, the PCR experiment demonstrated that the lesion didn't harbor the mutation of KRAS, BRAF, and PIK3CA which commonly presented in CRC through conventional carcinogenesis pathways. The present study discovered a panel which was composed of copy number alterations (CNAs) specific to the depressed neoplasm. The panel had a 66.7% accuracy, a 55.0% sensitivity, and an 84.6% specificity in the derivation cohort and a 79.7% accuracy, a 75.7% sensitivity and an 83.3% accuracy in the validation cohort. By using the panel to test 530 CRCs, 22.1% CRCs arise from the depressed neoplasms. The CRC arising from depressed neoplasm, so called de novo CRC, had a worse progression-free survival in comparison with conventional CRC. Moreover, the de novo CRC also has a worse progression-free survival in stage I, T1+2, and proximal cancer respectively. Conclusions FIT had a limited sensitivity for detecting SSA/P, and the use of IEE could increase the diagnostic yield of SSA/P by colonoscopy. The depressed neoplasm specific panel provides a window to develop a new diagnostic modality for this lesion. The de novo CRC had a unfavorable progression-free survival. Through using IEE for diagnosing SSA/P and developing the new tool for detecting depressed neoplasm may help to decrease the interval cancer from FIT false negative neoplasm.