Impact of follow-up interval on patients with hepatocellular carcinoma after curative ablation

An electronic medical record system, including a prospective follow-up database, has been maintained at Sun Yat-sen University Cancer Center since 2002 to track all cancer patients treated in the center, including HCC patients. In this prospective cohort study, all HCC patients who were initially treated with curative-intent thermal ablation (RFA/MWA) between January 2002 and January 2017 were identified. Ablation procedures were performed as described in our previous study [9]. A total of 856 patients were included using the following inclusion criteria: (a) Tumors within the Milan criteria; (b) Tumors exceeding the Milan criteria that were downgraded to within the Milan criteria by pre-ablation transcatheter arterial chemoembolization (pre-TACE); (c) No radiological evidence of major portal/hepatic vein branch invasion; (d) No extrahepatic metastasis; and (e) Child-Pugh A or B disease. Two hundred and forty patients were excluded based on the following exclusion criteria: (a) Patients without image (computed tomography (CT)/magnetic resonance imaging (MRI)) assessment of technique efficacy within 3 months after ablation; (b) Patients without complete ablation (complete ablation was defined as no enhancement in the ablated area on the first CT/MRI scan); and (c) Complete ablation achieved by multiple ablation courses. Finally, 616 patients were included in this study. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki, and the Ethics Committee of Sun Yat-sen University Cancer Center approved this study and written informed consent was obtained before treatment.

The first follow-up visit, approximately 1 month after ablation (3 months at most), was not regarded as the start of the surveillance regimen because it was performed to assess ablation efficacy. Follow-up visits were performed every 3–6 months until death or dropout from the surveillance program and each follow-up consisted of a physical examination, serum AFP test and at least one imaging examination (CT/MRI). To study the influence of surveillance interval on overall survival (OS), we took advantage of the known variability in patient compliance with the above follow-up strategy.

The patients were stratified into short- (< 4 months) or long-interval (4–6 months) groups by comparing the number of their actual follow-ups with the number of their expected follow-ups by the time of the last follow-up or when recurrence was detected [10]. As shown in the following example, one patient who had received 4 follow-ups within the 12 months after the first clinic visit was classified into the short-interval group because the actual number of follow-ups (4) was greater than the minimal expected number of follow-ups for the short-interval group (3). In contrast, one patient who had received 2 follow-ups within the 12 months after the first clinic visit was classified into the long-interval group because the actual number of follow-ups (2) was less than the maximal expected number of follow-ups for the long-interval group (3).

As all cirrhotic patients require ultrasound examination every 6 months [11, 12], it makes sense to recommend enhanced surveillance in patients after ablation for HCC. Hence, patients with surveillance intervals longer than 6 months (n = 171) were used only in the exploration of relapse risk factors, not in the surveillance analysis, as such scenarios were deemed irregular.

All consecutive post-ablation examinations performed for follow-up were retrieved from the electronic database. The recurrence data were used to explore the relapse pattern after ablation for HCC [13, 14]. From the RFS curve and the probability density plot, we found that more than half of patients (53.1%) experienced recurrence within 2 years after ablation, which is known as early recurrence. From the hazard rate curve, we found that the hazard of relapse reached its peak in the first year, decreased to its nadir in the third year and then increased again, which is considered late recurrence [15]. Late recurrences of HCC are thought to stem from de novo lesions related to the underlying chronic liver disease, persisting for the duration of the patient’s lifespan [11]. Thus, it makes sense to focus on surveillance during the first 2 years after ablation to detect early recurrence at a potentially more treatable stage. The characteristics of recurrent tumors and subsequent retreatments were also recorded. Curative-intent retreatments included resection, repeat ablation; other retreatments, such as TACE and biotherapy, were deemed as palliative [16].

Continuous variables were compared using the independent sample t-test and the Mann-Whitney U-test where appropriate. Binary and ordinal categorical variables were compared using the chi-squared test (Fisher’s exact test if necessary) and the Kruskal-Wallis test, respectively. OS curves were constructed and compared using the Kaplan-Meier method and log-rank test, respectively. All statistical analyses were performed using the R statistical package (R software version 3.3.2; R Foundation for Statistical Computing, Vienna, Austria) [17]. P values less than 0.05 were considered statistically significant, and all tests were two-tailed, except the tests for the BCLC stage of recurrent tumors, which were one-tailed because the stage of recurrent tumors in the short-interval group was not greater than that in the long-interval group.

We identified 616 patients who were initially treated with RFA/WMA of curative intent and the characteristics are shown in Table 1. The median follow-up period was 34.4 months (range 2–171 months). Overall, 319 (51.8%) patients developed recurrence, with 2-year RFS rate of 46.9%, and 5-year RFS rate of 19.3% (Fig. 1a), and 116 (11.8%) patients died, with 2-year OS rate of 91.8%, and 5-year OS rate of 72.7% (Fig. 1b). Relapse cases centered in the first 2 years after ablation (Fig. 1c), and recurrence hazard peaked in the first 2 years after treatment (Fig. 1d).

At multivariate analysis, tumor number (HR = 1.32; 95% CI, 1.00–1.73, P = 0.048) and tumor size (HR = 1.26; 95%CI, 1.11–1.43, P < 0.001; Table 2) were identified as risk factors for relapse, and accordingly, patients were classified into 2 groups with different relapse risks: the low-risk group (single tumor ≤3 cm; n = 318 patients) and the high-risk group (single tumor between 3 and 5 cm or 2–3 tumors ≤3 cm; n = 298 patients). The RFS curve of the high- and low-risk groups (HR = 1.98; 95% CI, 1.56 to 2.51, P < 0.001) was distinguished in Fig. 2.

Excluding the 171 patients with irregular surveillance, 445 patients with regular follow-up were classified into 2 groups: the short-interval group (interval < 4 months; n = 344 patients) and the long-interval group (interval between 4 and 6 months; n = 101 patients), and the mean surveillance interval was 2.58 months and 4.86 months, respectively. Finally, 199 (75.7%) and 64 (24.3%) patients underwent short- and long-interval surveillance in the low-risk group, and 145 (79.7%) and 37 (20.3%) patients underwent short- and long-interval surveillance in the high-risk group. The distribution of surveillance interval in the low- and high-risk groups was similar (P = 0.381), and the pertinent baseline clinic characteristics are shown in Table 3.

The 190 patients who underwent regular follow-up with early relapse were stratified into low- and high-risk groups. In the low-risk group, 54 (83%) patients with a short surveillance interval and 18 (72%) patients with a long surveillance interval were found to be relapsed with tumors at the BCLC 0/A stage (P = 0.172), and 44 (77.2%) patients and 18 (81.8%) patients received curative-intent retreatment, retrospectively (P = 0.886) (Table 4). However, in the high-risk group, patients in the short-interval group were found to have relapse more frequently at the BCLC 0/A stage than patients in the long-interval group (83% vs. 72%, P = 0.028), consequently, the proportion of patients who received curative-intent retreatment was higher for short-interval group (64.2% vs. 37.5%, P = 0.087), though with no significant difference (Table 4).

The 1-, 3-, 5-year OS of low-risk patients with short or long interval were 98.9, 91.2 80.4 and 100.0%, 89.1, 77.5%, respectively (P = 0.400) (Fig. 3a). The 1-, 3-, 5-year OS of high-risk patients with short or long interval were 99.2, 87.5, 69.9 and 97.1%, 72.7, 42.7%, respectively (P = 0.020) (Fig. 3b). Considering the 190 patients with early relapse, the 1-, 3-, 5-year OS of low-risk patients with short or long interval were 98.5, 81.1, 59.9 and 97.8%, 66.8, 36.3%, respectively (P = 0.025) (Fig. 3c).