In the present patient diagnosed as having intermediate-stage HCC, pathological complete response was confirmed by conversion surgery after neoadjuvant atezo/bev therapy. In the IMbrave 150 trial, the OS rate in the atezo/bev arm was significantly higher than that in the sorafenib arm. In addition, atezo/bev therapy was associated with a high overall response rate (ORR) of 44% in patients with intermediate-stage HCC and 27% in patients with advanced HCC as assessed according to RECIST 1.1, suggesting a tumor shrinkage effect of the treatment [5, 6].

Based on the results of this trial, Kudo argued that if atezo/bev is effective, it is important to consider conversion to curative therapy, such as resection or TACE, rather than just continuing atezo/bev therapy4). However, there have not been many case reports of conversion surgeries. [7,8,9,10,11,12,13,14,15,16].

The clinical CR rate in the atezo/bev group in the IMbrave 150 trial was 8%. However, 47.9% of patients in this trial had previously received local tumor treatments such as TACE. Kudo noted that pathological CR with systemic therapy alone is rare [4].

Cases of conversion surgery for HCC reported previously and our own case are shown in Table 1. Based on a review of these cases, we would like to discuss the merits and means of transition to conversion surgery.

Conversion therapy in patients with HCC offers the following advantages: (i) potential to achieve cancer-free and drug-free status; (ii) the possibility of evaluation of the pathologic response to chemotherapy; and (iii) the possibility of making future treatment choices, in the event of disease recurrence, based on the results of immunohistochemical analysis or cancer genome profiling. In our case, in terms of the treatment efficacy, liver resection performed as conversion surgery may have been unnecessary, because our postoperative histopathology revealed that complete tumor necrosis had already been achieved with the neoadjuvant atezo/bev therapy alone. Nevertheless, this effect of atezo/bev therapy would have been difficult to evaluate without surgery, as the tumor had still not disappeared on imaging.

A review of the reported cases of conversion surgery in the literature suggests that about 7 courses of atezo/bev, on average and median, were administered prior to surgery (Table 1). In our patient reported herein, we repeated the imaging during the atezo/bev therapy when the serum AFP level began to increase gradually, and found that fortunately, the tumor had not only shrunk in size, but was also resectable. Therefore, we performed conversion surgery. In retrospect, however, in our patient, even the CT performed after five cycles of atezo/bev therapy and one course of atezolizumab monotherapy showed that the tumor had become resectable, and we could have scheduled conversion surgery at that time point rather than continue atezolizumab monotherapy.

The optimal interval to conversion surgery after bevacizumab withdrawal, which has the side effect of delayed wound healing, remains controversial. Previous reports have recommended withdrawal of bevacizumab at least 5 to 8 weeks prior to the surgery, and resumption of the drug not earlier than 28 days after surgery or only after confirming complete wound healing [17,18,19].

Of the eight reported cases in which pathological CR was confirmed by conversion surgery, two showed leakage of contrast medium into the tumor, suggestive of intra-tumoral bleeding, during the course of atezo/bev therapy. It was not mentioned in the other 6 cases. Intra-tumoral bleeding causes tumor necrosis by inducing hypoxia, and a previous report has suggested that intra-tumoral hemorrhage is associated with a favorable prognosis in cases of HCC [20].

In addition, four of the previously reported cases of successful conversion surgery after atezo/bev treatment developed immune-related adverse events (irAEs) [7, 21, 22]. There are no reports of a statistical association between the development of irAEs and the efficacy of treatment against HCC yet, but such an association has been reported for other cancer types treated with immune checkpoint inhibitors (ICIs) [23, 24]. There are reports of existence of an association between the development of irAEs and treatment efficacy based on histopathological findings, but not in cases of HCC [25, 26].

Retrospectively, in our case, a very mild irAE may have occurred. AFP was gradually elevated, so surgery was performed, but the tumor was necrosis; when the reason for the elevated AFP was examined, AST and ALT were slightly increased by atezolizumab, as shown in Fig. 5. AST, ALT, and AFP normalized after postoperative withdrawal of the drug. In other words, the increase in AFP does not reflect the tumor but may have been re-elevated by chemotherapy-induced hepatitis. To determine if this hepatitis was irAE, the noncancerous areas of the resected liver were re-examined. The hepatitis was accompanied by a high degree of lymphocytic infiltration. This finding was not indicative of the presence of irAE, due in part to cirrhosis caused by hepatitis C (Fig. 7). Recent reports have demonstrated that PD-L1 is also expressed in normal hepatocytes and that atezolizumab causes hepatocellular necroptosis in human hepatocyte lines [27].

In addition, normalization of the serum levels of the tumor markers AFP and DCP was observed following atezo/bev therapy in all cases, except one that was not mentioned. Reduction of the tumor marker levels are indicative of a favorable tumor response, but normalization of the tumor marker levels is needed to achieve conversion surgery successfully.

In summary, in all the reported cases of successful conversion surgery, including ours, about seven courses of atezo/bev were administered until the patients were transitioned to conversion surgery. In cases of successful conversion surgery, irAE and normalization of serum tumor marker values often developed during the neoadjuvant atezo/bev treatment period, and intra-tumoral hemorrhage was observed in some cases although analysis of data from further cases is needed to confirm our findings.