Introduction

Ovarian high-grade serous carcinoma (OHGSC) is the most frequent histological subtype of ovarian cancer and the leading cause of death due to cancer of the female genital tract1. Typically, patients with OHGSC respond well to platinum-based chemotherapy, which is a standard regimen for OHGSC. However, OHGSCs frequently recur and gradually acquire resistance to standard chemotherapy regimens2. Recently, poly ADP-ribose polymerase (PARP) inhibitors have improved the clinical outcomes of patients with ovarian cancer featuring BRCA mutations or homologous recombination deficiency3. Nonetheless, more effective treatment strategies for advanced OHGSC are required. Immune checkpoint inhibitors targeting programmed death-1 (PD-1)/programmed death ligand-1 (PD-L1) and cytotoxic T-lymphocyte-associated protein 4 have caused breakthroughs in treatment strategies for various cancers4. In our previous study on ovarian cancer, PD-L1 expression was positively correlated with histone deacetylase (HDAC) 6 expression, and HDAC6 upregulation led to a poor prognosis5,6,7. HDAC6 increases deacetylated α-tubulin levels, which upregulate cancer cell growth by enhancing microtubule dynamics8,9. HDAC6 upregulation also promotes platinum resistance, and HDAC6 downregulation enhances platinum agent-induced DNA damage and apoptosis10. HDAC6 is also an immunomodulator, and the co-suppression of HDAC6 and PD-L1 has a synergistic anti-tumor effect for ovarian cancer11. Notably, Beltrame et al.12 and Takaya et al.13 have shown that the molecular profiles of OHGSC are significantly different before and after neoadjuvant chemotherapy (NAC). However, the evidence for HDAC6 and PD-L1 expression after NAC is not sufficient. In this study, we compared the expression of HDAC6 and PD-L1 by immunohistochemistry before and after chemotherapy to verify whether their expression affects chemotherapy resistance and patient prognosis in OHGSC.

Methods

Patients and samples

We identified 57 patients with OHGSC who were histologically diagnosed and had received NAC between 2007 and 2015 at Saitama Medical University International Medical Center. Clinicopathological, treatment, and follow-up data were also collected. The study protocol was approved by the institutional review board (IRB) (IRB number: 16–257) of Saitama Medical University International Medical Center. Written informed consent (or a formal waiver of consent) was obtained from all patients. Tumor samples were acquired before and after chemotherapy in all patients. Based on the results of an omental examination, chemotherapy response score (CRS) was used to classify patients as follows: patients with CRS3 had a complete/near-complete response, those with CRS2 had a partial response, and those with CRS1 had no response or a minimal response14,15. The surgical status was classified as complete resection (R0) or incomplete resection (R1). We also evaluate Response Evaluation Criteria in Solid Tumours (RECIST) and cancer antigen 125 ELIMination rate constant K (KELIM) score16.

Immunohistochemical staining and assessments

Immunohistochemical expressions of PD-L1 and HDAC6 were analyzed using formalin-fixed paraffin-embedded tumor samples from both before and after chemotherapy. A Dako Autostainer Link 48 (Agilent Technologies, CA, USA) was used according to the manufacturer’s protocol. The Target Retrieval Solution was applied for antigen retrieval at 98 °C for 20 min. Sections were incubated with primary antibodies (monoclonal rabbit anti-PD-L1, 1:100, 28 − 8 pharmDx, Dako North America, CA, USA; polyclonal rabbit anti-HDAC6, 1:500, ab1440, Abcam, Cambridge, UK) at 25 °C for 60 min, followed by incubation with a secondary antibody (EnVision FLEX/HRP, Agilent Technologies, CA, USA) at 25 °C for 30 min. The chromogenic reaction was performed using diaminobenzidine and hydrogen peroxide. Immunohistochemical expression was scored by two researchers (MiY and MaY) who were blinded to the clinicopathological characteristics (Fig. 1).

Fig. 1
figure 1

Immunohistochemical expression of histone deacetylase 6 (A before neoadjuvant chemotherapy [NAC]; B, after NAC) and programmed death ligand-1 (C before NAC; D after NAC).

Full size image

PD-L1 positivity was defined as staining in ≥ 5% of carcinoma cells17. PD-L1 expression was also analyzed in a semiquantitative manner by scoring the proportion of stained carcinoma cells over the total number of carcinoma cells, ranging from 5 to 100% in 5% increments. We also caluculated the mean of PD-L1 expression before and after NAC using sum of all PD-L1 percentages divided by the case numbers. High expression of HDAC6 was defined as staining in ≥ 50% of carcinoma cells6,7.

Statistical analyses

Statistical analyses were performed using Statistical Package for the Social Sciences (SPSS) version 24.0 (SPSS Inc., Chicago, IL, USA). Fisher’s exact test or Pearson’s chi-squared test was used to analyze the correlation between immunohistochemical expressions and clinicopathological characteristics. The Kaplan–Meier method was used to estimate survival curves, and the log-rank test was used to test differences between groups. The Cox proportional hazards model was used to perform multivariate survival analysis. P-values < 0.05 were considered statistically significant.

Results

Baseline characteristics and immunohistochemical expression after NAC

The baseline clinicopathological characteristics of the 57 patients with OHGSC who underwent platinum-based systemic chemotherapy as a neoadjuvant treatment are summarized in Table 1. The median follow-up period was 43.5 months (range 8–105 months).

Table 1 Patient background.
Full size table

Most patients (55 [96.5%]) underwent interval debulking surgery, but two (3.5%) patients only underwent a biopsy because of unresectable lesions. Of the 54 patients for whom omental CRS was able to be evaluated, 29 (53.7%) were classified as CRS3 (sensitive to chemotherapy), whereas 25 (46.4%) were classified as CRS1–2 (resistant or intermediate to chemotherapy). None of the patients received maintenance therapy with PARP inhibitors.

Change in PD-L1 and HDAC6 expression following NAC

Table 2 shows PD-L1 and HDAC6 expression in the 57 patients with paired pre- and post-NAC tumor samples. Before NAC, four patients showed high HDAC6 expression, and five patients showed positive PD-L1 expression. No significant correlations were shown between HDAC6 and PD-L1 expression after NAC.

Table 2 HDAC6 and PD-L1 expression before and after NAC.
Full size table

After NAC, high HDAC6 expression was observed in several patients (13, P = 0.019). The mean PD-L1-positive rate after NAC was 3.88%, which was significantly higher than the rate before NAC (0.68%) (P = 0.045). The association between patient characteristics and immunohistochemical HDAC6 and PD-L1 expression after NAC are summarized in Table 3.

Table 3 Correlation between clinicopathological characteristics and HDAC6 and PD-L1 expression.
Full size table

PD-L1-positive expression after NAC was significantly correlated with CRS3 (P = 0.042), which suggests a favorable response to chemotherapy.

Correlation between PD-L1 expression after NAC and clinical outcomes

Kaplan–Meier survival curves showed that positive PD-L1 expression and high HDAC6 expression after NAC were not significantly associated with progression-free survival (PFS) (P = 0.238 and P = 0.108, respectively) or overall survival (OS) (P = 0.493 and P = 0.377, respectively). Following multivariate analysis using the Cox proportional hazards model, surgical status was an independent prognostic factor for PFS (hazard ratio = 1.92; 95% confidence interval, 1.07–3.42; P = 0.028). No independent prognostic factors for OS were identified in the multivariate analysis.

We also performed a subgroup analysis based on the surgical status. Patients with R0 and PD-L1 positivity after NAC had poorer PFS (Fig. 2A) and poorer OS (Fig. 2B) than patients with R0 and PD-L1 negativity after NAC (P = 0.037 and P = 0.039, respectively).

Fig. 2
figure 2

Kaplan–Meier survival analysis (A PFS; B OS). Asterisks indicate the p-values for comparing R0 with PD-L1 negative to R0 with PD-L1 positive, and daggers indicate the p-values for R0 with PD-L1 positive to R1. P-values were calculated using the log-rank test. PFS progression-free survival, OS overall survival, PD-L1 programmed death ligand-1.

Full size image

However, no significant differences were observed in PFS or OS between patients with R0 and PD-L1 positivity after NAC and those with R1 (P = 0.971 and P = 0.705, respectively; Fig. 2A,B). HDAC6 expression after NAC was not significantly associated with PFS or OS in the subgroup analysis based on surgical status.

Discussion

The present study showed that immunohistochemical expression of HDAC6 and PD-L1 were upregulated in residual tumors after the initial platinum-based standard chemotherapy in OHGSC. Table 4 shows a review of studies regarding PD-L1 expression after NAC along with the prognosis for OHGSC17,18,19,20,21.

Table 4 Review of PD-L1 expression analysis before and after NAC for OHGSC.
Full size table

The first novelty of the present study is to report that high PD-L1 expression after NAC is significantly associated with poor PFS and OS in patients who undergo complete surgical resection. Upregulation of PD-L1 expression after NAC has been previously reported, but it has not been shown to contribute to the prognosis prediction. Guo et al. reported that elevated PD-L1 expression in residual tumors after platinum-based NAC was associated with a reduced chemotherapy response and inferior PFS in patients with lung cancer22. The mechanism underlying the role of chemotherapy in the variation in PD-L1 expression has not been fully elucidated. Peng et al. suggested that chemotherapy (platinum or taxane agents) could upregulate PD-L1 expression through nuclear factor kappa B to induce local immunosuppression in ovarian cancer23. Chemotherapy-induced immunosuppression may be a target for immunotherapy via PD1/PD-L1. A combination of chemotherapy (platinum and taxane agents) and pembrolizumab improved PFS and OS among patients with persistent, recurrent, or metastatic cervical cancer24. However, in patients with platinum-resistant ovarian cancer, nivolumab did not improve OS and PFS compared to gemcitabine or pegylated liposomal doxorubicin in a phase III study25. New therapeutic strategies are required to maximize the effects of immunotherapy targeting PD-1/PD-L1 in ovarian cancer. Notably, high expression of PD-L1 after NAC exhibits contradictory characteristics with good response to initial chemotherapy but unfavorable survival outcomes. Therefore, upregulated PD-L1 after NAC is a wolf in sheep’s clothing, and the physician should not make an optimistic prognosis even if the patient shows a good response to NAC.

The second novelty of the present study is to demonstrate that HDAC6 expression after NAC is upregulated compared to before NAC in OHGSC5. HDAC6 acts as an immunomodulator via PD-1/PD-L1, and co-suppression of HDAC6 and PD-L1 has a synergistic anti-tumor effect on ovarian cancer11. A phase Ib study of an HDAC6 inhibitor and nivolumab combination therapy for lung cancer has already been completed26. HDAC6 upregulation leads to resistance to platinum agents, whereas HDAC6 downregulation enhances platinum agent-induced DNA damage and apoptosis10. HDAC6 upregulation induces taxane resistance, and HDAC6 blockage reverses resistance to taxane agents via the acetylation of α-tubulin in ovarian cancer27,28. HDAC6 inhibitors exhibit an anti-tumor effect in vitro29,30,31,32, are well tolerated, and show minimal toxicity in phase Ib trials31. HDAC6 reduces kidney failure33 and peripheral neuropathy34, which are common adverse effects of platinum and taxane agents. Therefore, HDAC6 upregulation in residual tumors after NAC for OHGSC is considered a step in acquiring chemoresistance. HDAC6 suppression may lead to improved chemoresistance and synergistic effects with immunotherapy targeting PD-1/PD-L1.

This study has some limitations worth considering. First, the number of patients in the study may not have been sufficiently large. In particular, several subgroups, including the PD-L1-positive group, had only single-digit cases. The statistical legitimacy is debatable because of limited case numbers and multiple analyses. However, as Table 3 shows, this study is the third largest to date and can be interpreted as having a relatively large sample size for these types of studies. Second, the tumor immune environment, such as tumor-infiltrating lymphocytes and cluster of differentiation 8 immunohistochemistry, has not been investigated. However, this is a unique study that focuses on chemotherapy-induced changes in HDAC6 and PD-L1 expression. Finally, no molecular analysis was performed on the pre- and post-NAC specimens. We tried to analyze HDAC6 gene expression using published data, but we could not obtain the evidence of HDAC6 gene upregulations between pre- and post-NAC in OHGSC35,36,37. The molecular analysis will be the subject of future studies.

In conclusion, residual tumors of OHGSC after NAC show enhanced expression of HDAC6 and PD-L1, which are associated with tumor immunity, cell proliferation, and chemoresistance. PD-L1 expression also correlates with patient prognosis. These results suggest that HDAC6 and PD-L1 may be therapeutic targets and prognostic factors for residual tumors after standard chemotherapy in OHGSC.