Patients’ baseline characteristics
The clinical characteristics of 34 patients included in the present study are shown in Table 1. The median age was 73 years (range, 45–82 years). Regarding the cause of the background liver disease, 10 and 9 patients had hepatitis B and hepatitis C virus infections, respectively, and 15 patients had non-B, non-C hepatitis. Fifteen (44.1%) and 19 (55.9%) patients had Barcelona Clinic Liver Cancer classification stage B and C disease, respectively. Thirty-two (94.1%) and 2 (5.9%) patients were, respectively, classified as Child–Pugh class A and B. Eleven (32.4%) and 23 (67.6%) patients were classified as ALBI grades 1 and 2, respectively. Regarding classification based on the mALBI score, 20 patients (58.8%) were categorized as mALBI 1 and 2a (1/2a), and 14 (41.2%) as 2b. Of 34 patients, 13 (38.2%) had more than 10 intrahepatic tumors and 13 (38.2%) had extrahepatic metastases (i.e., nodal involvement and/or distant metastases) before the atezolizumab plus bevacizumab therapy. Twenty-four patients (70.6%) had a history of systemic chemotherapy, and the remaining 10 patients (29.4%) had not received HCC chemotherapy previously.
Changes in hepatic function and tumor markers
The changes in the proportions of ALBI scores at 1, 2, and 3 months (n = 34, 33, and 26, respectively) relative to the baseline scores are shown in Figure 1a. The median proportions of ALBI scores at 1, 2, and 3 months relative to the baseline scores were 0.93, 0.97, and 0.93, respectively. ALBI scores deteriorated at 1 month, slightly improved at 2 months, and deteriorated again at 3 months after the introduction of atezolizumab plus bevacizumab therapy. The changes in the ALBI scores according to the mALBI grade (1/2a versus 2b) are shown in Figure 1b. The same trend could be observed for improvements in liver function could be seen at 2 months for both groups. However, at 3 months after treatment, liver function improved and deteriorated relative to that at baseline in the mALBI 2b group (median ALBI score at 3 months relative to baseline score: 1.02) and mALBI 1/2a group (median ALBI score at 3 months relative to baseline score: 0.92), respectively; the difference was not significant. There were no significant differences in the changes in ALBI scores based on the treatment line (first-line versus second- or later-line) (Figure. 1c).
The changes in the proportions of AFP scores at 1, 2, and 3 months relative to the baseline scores are shown in Figure 2a (only for cases with preoperative elevation: AFP ≥ 10 [26/34]). The median proportions of AFP scores at 1, 2, and 3 months (n = 26, 22, and 20, respectively) relative to the baseline scores were 0.98, 1.12, and 1.83, respectively. In 13 of 25 (52%) patients, the AFP level decreased at 1 month compared to the baseline level. Of these 13 patients, 9 patients had evaluable levels at 3 months after the treatment, and in 2 of these patients, the AFP levels were higher than the corresponding baseline level at 3 months. In the remaining 7 patients, the AFP level gradually decreased for up to 3 months. Among the remaining 12 patients whose AFP level at 1 month was higher than that at baseline (proportion at 1 month relative to that at baseline: ≥1), 3 patients showed a decreased AFP level at 2 months compared to that at 1 month, while the remaining 9 patients showed increasing AFP levels with passage of time. There were no patients whose AFP level started to decrease at 3 months. The changes in AFP level according to the liver function determined based on the mALBI grade (1/2a versus 2b) and treatment line (first-line versus second- or later-line) are shown in Figures 2b and 2c. There were no significant differences in the changes in AFP score proportion according to both the liver function and treatment line.
Radiological evaluation and adverse events
Among the 34 patients, 32 (94.1%) could undergo radiological evaluation (Table 2). None achieved complete response, and 5, 25, and 2 patients showed partial response, stable disease, and progressive disease, respectively. The overall response rate (ORR) and disease control rate (DCR) were 15.6% and 93.7%, respectively. The ORRs were 10.5% (2/19) and 23.0% (3/13) for patients in the mALBI 1/2a and 2b group, respectively, and the corresponding DCRs were 89.5% (17/19) and 100% (13/13). There were no significant differences in ORRs and DCRs between the groups (P = 0.34 and 0.14). The ORRs were 33.3% (3/9) and 8.7% (2/23) for patients who received atezolizumab plus bevacizumab as first-line treatment and those who received this treatment as a second- or later-line therapy, respectively. The corresponding DCRs were 77.8% (7/9) and 100% (23/23), respectively, (p = 0.10 and 0.02). The ORR tended to be higher and DCR was significantly lower for patients in first-line group than that of those second- or later-line group (Table 2).
The most frequent adverse events associated with atezolizumab plus bevacizumab therapy at any grade was proteinuria, followed by serum aspartate aminotransferase / alanine aminotransferase disorder, decreased serum platelet count, general fatigue, and thyroid dysfunction, in this order. The proportion of adverse events of grade 3 or higher was 17.6% (6/34) and included perianal abscess (n = 1), proteinuria (n = 1), HCC rupture (n = 1), esophageal varices (n = 1), hypertension (n = 1), and perforation of the small intestine (n = 1).
Treatment course after atezolizumab plus bevacizumab therapy
Of the 34 patients who received atezolizumab plus bevacizumab therapy, 17 continued the regimen. Five patients discontinued chemotherapy and received the best supportive care, and 12 progressed to receive the next line of therapy. Of these 12 patients, all received lenvatinib as the next-line treatment option, and their clinical details are described in Table 3 (flowchart in Supplementary Fig .1). Ten patients started lenvatinib treatment within 14 days, and the remaining 2 patients started lenvatinib 21 and 24 days after the final administration of atezolizumab plus bevacizumab therapy, respectively. Among the 12 patients who received lenvatinib after atezolizumab plus bevacizumab therapy, 9 received a rechallenge with lenvatinib, and the remaining 3 had not received lenvatinib treatment previously. The changes in serum AFP level (only in those with an elevated AFP level: AFP ≥ 10 [8/12]) after lenvatinib introduction are shown in Fig. 3. All 8 patients who had elevated serum AFP levels before lenvatinib introduction showed a decrease in these levels in 1 month (Table 3). The median proportions of AFP levels at 1, 2, and 3 months relative to the baseline levels were 0.55 (0.31–0.94), 0.68 (0.56–3.07), and 0.98 (0.70–2.90), respectively. The AFP level increased in all 8 patients as time passed, from 1 to 2 months and 2 to 3 months, respectively, and it did not decrease in any patient.
The changes in ALBI scores after lenvatinib administration following the failure of atezolizumab plus bevacizumab therapy are shown in Figure 4. The median ALBI score was -2.24 (ranging from -3.11 to -1.68) before lenvatinib administration and -1.94 (range from -2.52 to -0.94) at 1 month after lenvatinib administration. Six of the 12 patients were classified into the mALBI grade 1/2a group before lenvatinib administration; however, 3 patients showed aggravation of liver function at 1 month, resulting in their classification into the 2b/3 group. The ALBI score worsened in 11 of the 12 patients, and the median proportion of ALBI score at 1 month relative to the baseline score was 0.81 (range: 0.46 to 1.18).