Analysis of Sorafenib Outcome: Focusing on the Clinical Course in Patients with Hepatocellular Carcinoma

Background Treatment outcomes of sorafenib therapy may greatly vary depending not only on tumor spread but also on past clinical processes prior to sorafenib therapy and timing of sorafenib administration in the past clinical course of hepatocellular carcinoma (HCC). We evaluated the efficacy of sorafenib in patients with HCC, taking into account of their past clinical courses. Methods Patients with HCC treated with sorafenib as a first-line systemic therapy, whose courses documented from the time of the initial diagnosis, were retrospectively analyzed. Results Of the 123 patients receiving sorafenib therapy at an advanced-stage, baseline characteristics differed including the rate of hepatitis C virus, Child–Pugh class, and status of intrahepatic lesions according to stage progression processes. Overall survival (OS) in patients progressed directly from the early-stage (15.3 months) was significantly longer than that in patients diagnosed at the advanced-stage (5.3 months, P = 0.022) and progressed from the intermediate-stages (6.0 months, P = 0.041). Of 105 patients diagnosed at the intermediate-stage on past clinical courses, OS of starting sorafenib therapy before progression to the advanced-stage (67 patients) was significantly longer than for patients starting sorafenib therapy only after progression to the advanced-stage (38 patients) (P = 0.015). Conclusion Characteristic differences between past stage progression processes might affect prognosis in advanced-stage HCC patients receiving sorafenib. Switching to sorafenib therapy before progression to the advanced-stage appears more effective than that after progression to the advanced-stage in patients diagnosed in the intermediate-stage on past clinical courses prior to sorafenib administration.

Introduction retrospective design. Patient records/information were anonymized and de-identified prior to analysis.

Patients
A total of 229 patients with HCC were consecutively treated with sorafenib as first-line systemic therapy between June 2009 and December 2014 at Chiba University Hospital, Japan. Among these, 39 patients were excluded due to one of the following reasons: (1) lack of data on the clinical process from the time of initial diagnosis (n = 27); (2) no TACE therapy after being diagnosed with intermediate-stage HCC, which is the standard therapy for intermediate-stage HCC (n = 12). Eventually, this study included 190 patients who received sorafenib as first-line systemic therapy and for whom data on the clinical process from the time of initial diagnosis were available, including liver function, tumor marker (alfa-fetoprotein [AFP]), and radiological assessment. For all patients, the presence of histologically confirmed or clinically diagnosed HCC could be documented (fulfilling the criteria for lesions with typical imaging) [6].

Sorafenib treatment
The treatment policy for patients with HCC followed the consensus-based clinical practice guideline proposed by the Japan Society of Hepatology (JSH) [4]. The timing of conversion from TACE to sorafenib was generally judged in the specialist's meeting based on the JSH guideline [4].
A total of 400 mg of sorafenib was orally administered twice per day (full dose = 800 mg). However, the initial dose of sorafenib was subject to reduction at the treating physician's discretion based on liver function and/or age. Sorafenib dose reductions (400 mg once daily or 400 mg on alternate days) and interruptions were allowed and depended on the type and severity of adverse events. Decisions on dose reduction and treatment discontinuation were informed by package insert data and experience with drug-related toxicities. We continued sorafenib administration until the development of intolerable toxicity or clear clinical disease progression. In our institution, the standard radiological follow-up procedure in sorafenib-treated patients was baseline, followed by evaluation in the first month after treatment, and then every 2 months. Because there is no satisfactory evidence in combination therapy, we did not combine sorafenib and other treatment such as TACE. If patients showed disease progression after sorafenib therapy, we considered whether definitive clinical trial could be accomplished, initially. All remaining cases are considered for best supportive care or sub-optimal treatments which were approved by the Ministry of Health, Labour and Welfare and were covered by social insurance of Japan, although there was no evidence-based treatment after sorafenib failure.

Clinical parameters
Clinical parameters included: baseline demographic data [gender, age, etiology, Eastern Cooperative Oncology Group performance status (ECOG-PS), Child-Pugh class, radiological assessment, AFP, treatment prior to the initiation of sorafenib therapy, and initial dose], adverse events after starting sorafenib, average daily dose, date of radiological progression according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 [16], progression pattern, liver function deterioration at the end of sorafenib treatment, post-sorafenib therapy, and date of death or last follow-up. Data were collected via the database of sorafenib-treated patients in our institution. We also reviewed medical records to identify the history of the clinical course of HCC prior to sorafenib administration and the initial diagnosis of each patient (i.e., early-, intermediate-or advanced-stage HCC). We then adjusted the clinical parameters at the time of diagnosis at the intermediate-stage (age, Child-Pugh class, radiological assessment, and AFP) and effectiveness of initial TACE in the intermediate-stage according to the modified RECIST (mRECIST) [17]. Patients responding to initial TACE were defined as patients who had achieved a partial or a complete response for more than 3 months after TACE. Data on ECOG-PS for patients diagnosed with intermediate-stage were removed from the final data set available for analysis because they could not be identified objectively from the medical charts.
In this study, we defined each HCC stage as follows: (1) early-stage: single lesion of any size or 2-3 lesions of 30 mm diameter; (2) intermediate-stage: >3 lesions of any size or 2-3 lesions of >30 mm diameter; (3) advanced-stage: macrovascular invasion (MVI) or extrahepatic metastasis (EHM). Stage progression processes of patients starting sorafenib therapy at the advanced-stage were classified into three categories (advanced-stage at the time of initial diagnosis, progression to advanced-stage through the intermediate-stage, progression to the advanced-stage directly from the early-stage). Intermediate-stage patients were also classified into two groups as follows: those who had already developed intermediate-stage HCC upon initial diagnosis and those who had progressed from the early-stage.

Statistical analysis
The chi-square test or Fisher's exact test was used to compare demographic and clinical characteristics as appropriate. Kaplan-Meier plots of medians [with 95% confidence intervals (95% CI)] were used to estimate overall survival (OS). The censoring date was defined as the date of the last follow-up. Time to progression after sorafenib administration was estimated using Kaplan-Meier plots of medians (with 95% CI), with the censoring date being defined as the date of the last radiological assessment without progression. Univariate and multivariate Cox proportional-hazard models were used to estimate the hazard ratios for risk factors predicting OS. A probability (P) value of <0.05 was considered statistically significant. We defined a cutoff value of maximum size (50 mm/>50 mm) and number (7/>7) of intrahepatic lesions based on our previous study [11]. Propensity scores were used to generate matched pairs for comparisons of survival between the timing of sorafenib administration before and after progression to the advanced stage in patients with intermediate-stage HCC. Possible variables associated with survival of the patients with intermediate stage HCC, including tumor number, maximum size of the tumor, Child-Pugh class, AFP, and response to initial TACE, were comprehensively selected for propensity score generation. Logistic regression analysis with the selected variables was used to generate a continuous propensity score from 0 to 1. A one-toone nearest neighbourhood match between the two groups was used to select patients for subsequent analysis. All statistical analyses were performed using SPSS statistical software (version 23; SPSS-IBM, Chicago, IL, USA). Table 1 displays the baseline characteristics of 190 enrolled patients with HCC. Most patients were males (77%), and the median age was 72 years (range, . The most frequent etiology was HCV (50%), followed by HBV (15%), and alcohol abuse (9%). One hundred thirty-eight patients (73%) were classified as Child-Pugh A, and 52 patients (27%) were classified as Child-Pugh B. Sixty-seven patients (35%) had MVI and 85 patients (45%) had EHM.

Sorafenib therapy in advanced-stage HCC patients
The baseline characteristics of patients starting on sorafenib therapy at advanced-stage HCC are included in S1 Table. Patients directly progressing from early-stage HCC had the highest rate of HCV infection (69%) compared with the proportion of patients diagnosed with the advanced-stage (25%) and those who progressed through the intermediate-stage (45%) (P < 0.001). The majority (75%) of patients initially diagnosed with advanced-stage HCC had intrahepatic MVI. Meanwhile, 31% of patients that directly progressed from the early-stage did not have intrahepatic lesions. Hence, the status of intrahepatic lesions significantly differed among the three groups (P < 0.001).
The median OS in advanced-stage HCC patients was 7.6 months (95% CI: 5.4-9.8). Kaplan-Meier estimates of OS according to stage progression process are shown in Fig 2. The OS of patients initially diagnosed with advanced-stage HCC, progressing through the intermediate-stage, and directly progressing from the early-stage was 5.3 (95% CI: 3.9-6.7), 6.0 (95% CI: 1.3-10.7), and 15.3 months (95% CI: 12.2-18.5), respectively. Although there was no significant difference between patients initially diagnosed with advanced-stage and those who progressed through the intermediate-stage (P = 0.822), patients directly progressing from the early-stage HCC experienced significantly longer survival compared with the other two groups (vs. patients initially diagnosed with advanced-stage HCC: P = 0.022; vs. patients progressing through the intermediate-stage: P = 0.041). Multivariate analysis identified baseline age, ECOG-PS, status of intrahepatic lesions (none, without MVI, and with MVI), number of intrahepatic lesions as independent predictors of OS in patients with advanced-stage HCC ( Table 2; univariate analysis are shown in S2 Table). Meanwhile, stage progression process had no predictive power in this analysis.

Sorafenib therapy in intermediate-stage HCC patients
The baseline characteristics of patients starting on sorafenib at the intermediate-stage are included in S3  (Table 3; univariate analysis are shown in S4 Table).
Progression pattern of sorafenib treatment, liver function deterioration at the end of sorafenib treatment, and post-sorafenib therapy S5 Table shows correlations of the OS and progression pattern of sorafenib according to previous reports [18][19][20], with liver function deterioration at the end of sorafenib treatment defined     1-10)]. Upon analysis of demographic data and other characteristics for these patients, no significant difference was observed between starting sorafenib before or after progressing to advanced-stage HCC, including sub-classification model of intermediate-stage [21]. Of 67 patients who started sorafenib before progressing to advanced-stage HCC, 64 patients (96%) were deemed as TACE failure/refractory according to the JSH guideline before conversion from TACE to sorafenib [4]. Of 38 patients who started sorafenib after progressing to advanced-stage HCC, 24 patients (63%) deemed as TACE failure/refractory without MVI and EHM. These 24 patients continued TACE even if they deemed TACE failure/refractory and started sorafenib after progressing to advanced-stage HCC. The remaining 14 patients (37%) were deemed TACE failure/refractory with either or both of MVI and EHM. The median time from the day of last TACE to advanced-stage HCC was 3.2 months (95%CI: 2.8-3.6). The median survival from the day of the last TACE was significantly longer than in patients who started sorafenib before progressing to advanced-stage HCC than in those who started sorafenib after progressing to advanced-stage HCC . Although selection bias between two groups might exist, we also created propensity score matched dataset (S6 Table). Multivariate analysis of both any patient and propensity matched dataset identified the timing of sorafenib administration as an independent predictor of OS in patients with intermediate-stage HCC, as well as the effectiveness of initial TACE in intermediate-stage HCC ( Table 5; univariate analysis are shown in S7 Table).

Discussion
Focusing on the clinical course of HCC until the administration of sorafenib, we attempted to organize the heterogeneous population of sorafenib-treated patients. This approach differed from previous studies that have evaluated the outcome of sorafenib-treated HCC patients in field practice [22][23][24][25][26] and might give us a new perspective.  Our study showed that the baseline characteristics in sorafenib-treated patients with advanced-stage HCC differed among patient groups with different stage progression processes. Patients initially diagnosed with advanced-stage HCC had significantly high rates of i) Child-Pugh B class, ii) intrahepatic lesions with MVI, and iii) absence of HCV infection. The majority of these patients, particularly patients in whom HCC is not associated with virus infection, might not have been subjected to HCC screening. Thus, they were diagnosed with advanced-stage HCC with highly progressing and decreasing liver function. On the other hand, we have often experienced HCC patients who recur with EHM but with or without only minor intrahepatic lesions upon curative therapy. Of importance, these patients were not complicated by MVI in many cases. Our results confirmed that 62% of patients directly progressing from the early-stage disease either did not have intrahepatic lesions (31%) or had intrahepatic lesions without MVI (31%). Progression through the intermediate-stage to the advanced-stage is a common clinical course in HCC. This study showed that intrahepatic lesions were more frequently accompanied by MVI than that of the other two groups. Moreover, the rate of HCV and Child-Pugh B class were between the other two categories. Although all of them were classified as having advancedstage HCC treated with sorafenib, they were regarded as clinically different categories. Based on a meta-analysis of four randomized control trial, Shao YY reported that sorafenib might provide survival benefits to patients positive for HCV [27]. However, our results might indicate that differences in the rate of HCV between stage progression processes influenced that result.
Next, we performed separate analyses of the prognosis of sorafenib therapy in intermediateand advanced-stage patients. The results of the multivariate analysis of patients with advancedstage HCC demonstrated that the status of intrahepatic lesions was an independent prognostic factor. Several studies have indicated that the status of intrahepatic lesions and/or MVI were significant predictors of survival in patients with EHM [28,29]. Sohn et al. reported that patients characterized by intrahepatic lesions with MVI experienced shorter survival times compared with patients characterized by intrahepatic lesions without MVI, and patients harboring intrahepatic lesions in sorafenib treated patients with EHM [23]. Considering these findings, status of intrahepatic lesions, particularly the presence of MVI, were strong, poor predictive factors for sorafenib-treated patients with advanced-stage HCC. Stage progression process from initial diagnosis appeared to be associated with the status of intrahepatic lesions in advanced-stage HCC. Our results indicated that the OS of patients directly progressing from early-stage HCC was significantly longer than that observed for the other two categories. Although stage progression process was not an independent prognosis factor, we believe that the differences in clinical characteristics among the three categories, particularly the status of intrahepatic lesions, might affect patient prognosis.
In intermediate-stage patients who started sorafenib, our results identified AFP as the only independent prognostic factor. Previous studies have shown that the maximum size and/or number of intrahepatic lesions are predictive factors of intermediate-stage HCC patients who have received TACE [11,30,31]. Thus, the prognosis of intermediate-stage patients is thought to depend on these two factors as well as liver function [21,32]. In contrast, our finding that both maximum tumor size and number were not predictive factors of survival in sorafenibtreated patients. This might be relevant for considering treatment strategies for intermediatestage HCC patients. Prior to this study, there have only been a few studies that evaluated the outcome of sorafenib therapy, specifically with regard to intermediate-stage patients [18,33], and this is the first report to analyze prognostic factors of sorafenib therapy in intermediatestage HCC patients. In this analysis, ECOG-PS and Child-Pugh class were not independent prognostic factors in spite of being significant factors of the univariate analysis. These factors have been known to contribute to the survival after sorafenib therapy [22,34]. Although our result was from a small number and a single institution, additional analyses should be conducted to confirm the outcome of sorafenib treatment in patients with intermediate-stage HCC.
This study identified that switching to sorafenib before progression to advanced-stage disease conferred a survival benefit compared with switching to sorafenib only after progression to advanced-stage disease in patients diagnosed in intermediate-stage HCC on the past clinical course prior to sorafenib therapy. Although this analysis was based on retrospective data and it might have selection bias and lead time bias of conversion from TACE to sorafenib, our results indicate that the difference in timing of switching to sorafenib was associated with the outcome of the two groups. After the approval of sorafenib, several suggestions for alternative therapies in patients with TACE failure/refractoriness have been made [4,14,15]. Moreover, two reports demonstrated that switching to sorafenib upon TACE failure according to the definition of TACE failure/refractoriness conferred longer survival compared with continued TACE in patients with intermediate-stage HCC [35,36]. These findings, coupled with our results, could indicate that switching to sorafenib from TACE in intermediate-stage HCC patients is an effective treatment strategy for patients diagnosed with intermediate-stage HCC compared with continuing TACE until progression to advanced-stage HCC.
For major solid cancers, decisions on treatment often rely simply on whether they are "resectable" or "unresectable," and first-line systemic therapies are initial treatments for "unresectable" patients (NCCN Gudelines 1 ; http://www.nccn.org). With HCC, the heterogeneous clinical courses and treatment processes associated with the disease complicate the picture as follows compared with major solid cancers. First, HCC has a high potential of recurrence because of primary tumor spread and persisting hepatocarcinogenesis [9]. Second, there are several treatment options for "unresectable" patients such as local ablation, TACE, and sorafenib therapy [4][5][6]. Patients typically receive treatments that are considered suitable at that particular time of HCC progression. Finally, liver function is a strong prognostic factor and needs to be considered before making treatment choices [4][5][6]. Because it is not feasible to use sorafenib as an initial treatment option in most patients [22,23,25,34], the clinical course before starting sorafenib and analyzing the outcome of sorafenib therapy in heterogeneous populations should be considered.
In conclusion, the characteristics of patients receiving sorafenib were different according to stage progression process from the time of initial diagnosis. These factors might affect the prognosis of sorafenib therapy in advanced-stage HCC patients receiving sorafenib. Switching to sorafenib from TACE before progression to advanced-stage HCC appears to be effective compared with switching to sorafenib only after progression to the advanced-stage. Additional, large-scale studies that take into account the clinical course of HCC are needed to identify the suitable timing of sorafenib administration over the clinical course of HCC.
Supporting Information S1 Table. Baseline demographic data of patient characteristics at the time of starting sorafenib (advanced-stage hepatocellular carcinoma patients).