Oncotarget. 2017;8(63):106499-106510.

Development of a nomogram for screening of hepatitis B virus-associated hepatocellular carcinoma.

, , , , , , ,

제1저자 정정화 · 교신저자 김진욱

PMID 29290966 PMC5739751 DOI 10.18632/oncotarget.22498 IF 5.168 소화기내과

소속 (3)

Department of Medicine, Seoul National University Bundang Hospital, Seongnam, Republic of Korea.
Department of Internal Medicine, Seoul National University College of Medicine, Seoul, Republic of Korea.
Department of Radiology, Seoul National University College of Medicine, Seoul, Republic of Korea.

Abstract

Current strategy of hepatocellular carcinoma (HCC) surveillance evaluates individual risks of HCC for defining candidates for surveillance, but estimated risks are not utilized for clinical decision-making during actual screening. We sought to determine whether consideration of individual risks improve the performance of ultrasound (US)-based HCC screening in a real-world chronic hepatitis B (CHB) cohort. This single center retrospective cohort study analyzed 27,722 screening US tests from 4,175 consecutive CHB patients. Logistic regression analysis was performed to identify independent parameters predicting presence of HCC. A nomogram was built based on the independent predictors of HCC and compared with US-only screening by receiver operating characteristics analysis. The cost-effectiveness of the nomogram was assessed by decision curve analysis. HCC developed in 222 patients with the incidence of 0.769 per 1000 person-year during the median follow-up of 63 months. Age, sex, presence of cirrhosis, serum alpha-fetoprotein (AFP) levels and positive US test results were independent predictors of HCC presence. A nomogram based on these predictors showed higher C-statistics compared to US-only screening (0.960 vs. 0.731 and 0.935 vs. 0.691 for derivation and validation cohort, respectively; p < 0.001). Decision curve analysis showed higher net benefit of the HCC nomogram-guided screening model compared to US-only screening in the risk threshold range between 0 and 0.3. A nomogram composed of age, sex, presence of cirrhosis, serum AFP levels and US findings better predicted the presence of HCC compared to US-only screening in CHB on surveillance.

Keywords

chronic hepatitis Bearly diagnosis of cancerhepatocellular carcinomanomogramsultrasonography

Figures

**Figure 2** Nomogram for predicting presence of HCC in chronic hepatitis B patients on surveillance The individual point score for each variable is obtained on the corresponding perpendicular position on the top “Points” axis. Continuous values, i.e. age and logAFP, outside of the boundaries are replaced by the corresponding boundary value. The sum of all points, HCC nomogram scores, are converted to predicted HCC probability on the bottom probability axis.

**Figure 3** Calibration of the HCC nomogram score model The predicted probability of HCC presence was plotted against HCC nomogram score in the derivation dataset (A) Agreement between the predicted and observed HCC probabilities were plotted for the derivation (B) and validation (C) datasets with 300 bootstraps. The HCC nomogram score showed good calibration within the expected probability range up to 0.3, which corresponds to HCC nomogram score of 195. Hosmer and Lemeshowʼs goodness-of-fit test showed no significant discrepancies between the predicted and observed probabilities for HCC presence ( p = 0.72 and 0.82 for derivation and validation set, respectively).

**Figure 4** Decision curve analysis for HCC screening models Decision curves for derivation set (A) and validation set (B). Ultrasound-only indicates traditional US-based screening in which decisions to request confirmatory tests are guided only by positive US tests. *HCC nomogram* indicates that the decisions are guided by HCC nomogram scores. Risk threshold and cost:benefit ratio indicate the relative significance of correct detection of HCC to correct exclusion of HCC of the models. *All* indicates that all CHB patients receive confirmatory tests, i.e. dynamic imaging studies or biopsy, and *None* indicates that no patients receive confirmatory tests. The net benefit of HCC nomogram was higher than that of US-only across given range of threshold probabilities, indicating that nomogram-based screening model would produce cost-effective clinical outcome irrespective of patient preference.

그림 출처: PubMed Central 공개 논문 (PMC5739751)

Conflict of interest statement

CONFLICTS OF INTEREST The authors have nothing to declare.

그 밖의 정보