危險分數分層導向之肝癌超音波篩檢效益評估

Abstract

背景：肝癌是目前台灣地區惡性腫瘤死亡的第一位，就發生率而言亦為好發腫瘤之一，肝癌的篩檢為目前防治肝癌的必要手段。本論文主要目的在早期評估以危險分數為導向的肝癌社區到點超音波篩檢計畫之效益。 材料和方法：本研究的對象取自2005至2010年間參加彰化縣社區整合式篩檢且年齡介於45-69歲間共計41412人。利用已發展的危險分數公式將參加者區別為極高危險群、高危險群、中危險群和低危險群。個案中符合HBsAg(+)、Anti-HCV(+)、異常AFP中的其中一項，利用危險分數計算公式，並以危險分數的中位數區隔為極高危險群及高危險群。其他三項指標皆陰性者，亦利用危險分數計算公式得到其危險分數，再以前1/4及後3/4區分為中危險群和低危險群。並由衛生局主導,依危險分數作為優先性，並邀請個案至衛生所接受肝癌超音波檢查。利用比例風險回歸模型評估超音波篩檢偵測、兩階段篩檢偵測及臨床症狀偵測三類個案之風險，並調整前導期偏差。利用盛行池概念計算不同危險分層個案之平均滯留期，並推估不同危險分層在不同篩檢間隔之下的篩檢效益。 結果：本計畫經由超音波篩檢偵測之確診肝癌共19位。在極高危險群有16位肝癌，偵測率為千分之5.3 (=16/3013)；在高危險群為千分之0.7 (=2/2732)；在中間危險群為千分之0.4 (=1/2819)，而在低危險群中接受超音波檢查之426人並沒有發現肝癌個案。19位肝癌中，腫瘤大小小於3公分為8位，佔42.11%；3-5公分為9位，佔47.37%；大於5公分為2位，佔10.53%。超音波偵測、兩階段篩檢偵測及臨床偵測個案的之危險比，分別為0.15 (95% CI: 0.04-0.59)及0.41 (95% CI: 0.21-0.78)，前導期偏差調整後危險比分別提高為0.31 (95% CI:0.12-0.84)及0.63 (95% CI: 0.37-1.08)。極高危險群、高危險群、中危險群之平均滯留期為 1.10年、1.27年、3.55年。 就極高危險群而言，每季篩檢一次約可減少61%肝癌死亡，每半年一次篩檢其篩檢效益約為54%，若為每年，則可減少約51%，此效益隨著篩檢間隔變長而變小，若為每六年一次篩檢，則僅約減少14%肝癌死亡。在其他危險分層亦可見相似情況。若以高危險群每半年篩檢一次，即目前台灣地區全民健康保險的策略為基準，欲避免一個肝癌死亡需進行37,547次腹部超音波篩檢，此效益類推至中危險群可建議此類族群的篩檢間隔約介於4-6年，但對低危險群而言，就算六年一次篩檢，所需使用的超音波檢查資源亦高於高危險群每半年篩檢一次的結果。 結論：以社區為基礎、危險分數為導向的腹部超音波篩檢計畫可以找到較多小型肝癌，並得到較佳的存活狀況。依據危險分數決定優先性的邀約及以衛生所為定點提供的到點篩檢服務可增加民眾參與率，並提高篩檢產出。本計劃的結果有助於未來危險分層篩檢之政策推行。

Background: Community-based ultrasound screening for HCC is one of approaches to the prevention of hepatocellular carcinoma (HCC) for adults aged 30 years or older who had not the chance of receiving vaccination against hepatitis B virus infection. Weaimed to evaluate the mass screening for HCC with ultrasound in Changhua, Taiwan, one of areas with higher incidence of HCC. Method: A total of 41412 residents aged 45-69 years attending the ChangHua Community-based Integrated Screening (CHCIS) constituted our study population. All participants were invited to screen in the light of four risk groups: extremely high (EH), high (H), intermediate (IM), and low (L). EH and H were classified by the median of risk score among those with positive HbsAg or anti-HCV or elevated alpha-fetoprotein level. IM and L were classified by the third quartile of the risk score among those with neither positive of the three markers. Participants were invited to an out-reaching screening program for HCC with ultrasound. We compared survival curves for those detected by ultrasound screening (ultrasound-detected), detected by two-stage method (two-stage screen-detected), and those diagnosed due to the appearance of clinical symptoms (clinically-detected). Proportional hazards regression model was used to assess the hazard ratios of risk for death from HCC by detection modes. We also calculated the adjusted hazard ratio considering lead-time bias. Prevalence pool was used to estimate the mean sojourn time (MST) in the four risk groups. Efficacy in terms of number-needed-to-screening (NNS) to avoid one interval cancer and NNS to avoid one HCC death was reported. Results: A total of 19 HCC were confirmed after ultrasound screening. Among them, 42.11% (n=8) had tumor size less than 3 cm, 47.3% (n=9) in 3-5 cm, and 10.53% (n=2) cases had tumor larger than 5 cm. Among the 19 ultrasound-detected cases (median follow-up time: 21.6 months), 2 (10.53%) died from HCC at 24 and 25 months. There were 24 two-stage screen-detected cases (median follow-up time: 37.3 months), of which 9 (37.5%%) died from HCC. Another 1165 clinically-detected were also identified, of which 798 (68.5%) died from HCC in a median follow-up time of 15.3 months. Compared with clinically-detected cases, the hazard ratios of HCC death were 0.15 (95% CI: 0.04-0.59) and 0.41 (95% CI: 0.21-0.78) for ultrasound-detected and two-stage screen-detected,respectively. The corresponding figures for lead-time bias adjusted hazard ratio was 0.31 (95% CI:0.12-0.84) and 0.63(95% CI 0.37-1.08), respectively. The mean sojourn time (MST) for EH, H, and IM were estimated as 1.10, 1.7, and 3.55 years, respectively. Three-monthly ultrasound screening could reduce 61% mortality from HCC for the EH group. The efficacy was reduced to 54% for 6-monthly screening, and 51% for 1-yearly screening. It was further reduced when the interscreening interval was further lengthened. Applying the current policy from the government, half-yearly ultrasound check-up for hepatitis B or C virus infection, it requires 4,383 and 37,547 ultrasounds to avoid one HCC death for EH and H groups. Conclusion: Ultrasound screening can detect small hepatocellular carcinoma with favor survival even after adjustment for lead-time bias. The out-reaching can also enhance yield of screening, even for those without positive for HBsAg or anti-HCV. The results of this program facilitate the policy for risk-stratification tailored screening in the future.