Changes in HER3 expression profiles between primary and recurrent gynecological cancers

This study included gynecological cancer patients with matched-pair tissue samples taken at initial diagnosis and at recurrence between 1999 and 2019 at the National Cancer Center Hospital, Japan. Patients with unavailable or insufficient tumor tissue were excluded. Finally, 86 patients were included (40 with ovarian, 32 with endometrial, and 14 with cervical cancers; Additional file 1: Table S1, Additional file 2: Table S2, Additional file 3: Table S3). We retrospectively collected the following clinical and pathological data: age, histology, stage as defined by the International Federation of Gynecology and Obstetrics in 2008 [33], lymph node metastasis, adjuvant treatment, and postoperative survival time.

The study was approved by the Institutional Review Board of the National Cancer Center, Tokyo, Japan (Approval Number: 2020–003). Written informed consent was waived because of the retrospective design.

We performed immunohistochemical (IHC) staining for HER3 as previously described [34, 35]. Briefly, sections of each sample were deparaffinized and antigen retrieval was performed at high pH (PT Link machine, Dako). The sections were then stained with a rabbit monoclonal antibody against HER3/ErbB3 (1:59 dilution; clone D22C5, Cell Signaling Technology Inc., Danvers, MA, USA) using the Dako autostainer Link48 (Dako, CA, USA) and EnVision Flex Mini Kit (Dako), according to the manufacturer’s instructions. Hematoxylin was used as a nuclear counterstain. HER3-high was defined as an IHC score of 3 + or 2 + , and HER3-low/zero was defined as an IHC score of 1 + or 0 in line with the HER2 testing guidelines for gastroesophageal cancer [36]. The H-score (range, 0–300) was calculated using the following formula: 3X + 2Y + Z, where X, Y, and Z are the percentages of tumor cells showing strong, moderate, and weak staining intensities, respectively [37] (Additional file 1: Table S1A–D). The specificity of this HER3 IHC staining method was validated by confirming no signal detected with an isotype control antibody. Both positive and negative cell line control slides were also incorporated into HER3 staining to verify the specificity at every batch of staining.

Differential HER3 expression between initial diagnosis and at recurrence was evaluated using the following tests; the χ² test for categorical variables and the Mann–Whitney U test for continuous variables. Overall survival (OS) was estimated using the Kaplan–Meier method. The log-rank test was used to compare survival between groups. OS was defined as the time from relapse to death from any cause. All tests were two-tailed, and the significance level was set at α = 0.05. All statistical analyses were performed using GraphPad Prism ver.8.0 (GraphPad Software, San Diego, California, USA).

HER3 expression IHC scores at initial diagnosis were ≥ 1 + in 95.0% of cases, while 67.5% (27 cases) showed scores ≥ 2 + (HER3-high; Fig. 1a and Additional file 4: Table S4). At recurrence, 80.0% (32 cases) had HER3-high scores (Fig. 1a, b). H-scores were significantly higher at recurrence (p = 0.0041, t-test; Fig. 1c); 27.5% of patients had altered HER3 designations (Fig. 1d). Of the 13 HER3-low/zero patients, 61.5% (eight cases) changed to HER3-high. Among the 27 HER3-high patients, 11.1% (three cases) changed to HER3-low/zero.

No significant differences were found for HER3-high scores at recurrence in any of the following: high-grade serous carcinomas vs. others (77.8% and 84.6%, respectively), platinum-sensitive recurrence vs. -resistant recurrence (79.3 and 81.8%, respectively), localized relapse vs. distant metastases (75.0 and 83.3%, respectively), and one type of prior therapy vs. two or more (78.5 and 83.3%, respectively; Additional file 7: Figure S2a–d). In samples from patients receiving two or more chemotherapeutic regimens, the HER3-high score changed significantly between initial diagnosis and at recurrence (p = 0.04; Additional file 7: Figure S2d).

Of the patient, 46.9% (15 cases) with endometrial cancer were HER3-high at initial diagnosis (Fig. 2a and Additional file 5: Table S5). At recurrence, 68.8% (22 cases) were HER3-high (Fig. 2a, b). H-scores tended to be lower at initial diagnosis (median, 65) than at recurrence (median, 100); however, this difference was not significant (p = 0.08, t-test; Fig. 2c). Changed HER3 score were identified in 53.1% (17 cases) of endometrial cases (Fig. 2d); 70.6% of patients (12 cases) changed from HER3-low/zero to HER3-high category, while 33.3% (five cases) changed from a HER3-high to a HER3-low/zero score.

Low-grade endometrial cancer was defined as G1 and G2, and high-grade endometrial cancer was denoted as others; 13.3% (two cases) of samples from low-grade endometrial cancers were HER3-high at initial diagnosis, which increased significantly to 86.7% (13 cases) at recurrence (p < 0.0001; Additional file 7: Figure S2e). No significant differences were found for HER3-high scores at recurrence in any of the following: localized relapse vs. distant metastases (66.7% and 72.7%, respectively) and non-chemotherapy vs. prior chemotherapy (61.5% and 73.7%, respectively) (Additional file 7: Figure S2f, g). In patient samples with distant metastases or prior chemotherapy, the HER3-high score increased significantly between initial diagnosis and at recurrence (p = 0.01 and p = 0.049, respectively; Additional file 7: Figure S2f, g).

Of the patient population, 85.7% (12 patients) with cervical cancer were HER3-high at initial diagnosis (Fig. 3a and Additional file 6: Table S6); 85.7% (12 patients) were also HER3-high at recurrence (Fig. 3a, b). H-scores for HER3 showed a tendency to increase from initial diagnosis (median, 122.5) to recurrence (median,180), but this change was not significant (p = 0.19, t-test; Fig. 3c). HER3 expression status changed in 14.3% (two cases) of patients. One was HER3-high at diagnosis and HER3-low/zero at recurrence, while the other was HER3-low/zero at diagnosis and HER3-high at recurrence.

No significant differences were found for HER3-high scores at recurrence in any of the following: squamous vs. non-squamous (75.0% and 100.0%, respectively), localized relapse vs. distant metastases (87.5% and 83.3%, respectively), and non-radiotherapy vs. prior radiotherapy (83.3% and 87.5%, respectively).

We analyzed survival in patients with ovarian and endometrial cancers, excluding cervical cancer, which is much less common. Carcinosarcoma was excluded from the analysis among endometrial cancers because of its poor prognosis. At recurrence, the median survival time for HER3-high ovarian and endometrial cancers tended to show poorer OS than HER3-low/zero cancers. In ovarian cancers, HER3-low/zero cancers had an unreached OS (95% confidence interval (CI), 4.4 to 73.3 months), while HER3-high cancers had an OS of 40.9 months (95% CI 12.5 to 30.1 months), with a hazard ratio (HR) of 0.41 and a 95% CI of 0.13–1.24. In endometrial cancers, HER3-low/zero OS was 82.2 months. (95% CI 6.5 to 79.2 months), while HER3-high cancers had an OS of 28.1 months (95% CI 17.0 to 33.5 months), with an HR of 0.65 and 95% CI of 0.22–1.89 (Fig. 4a, b). In addition, we examined survival based on changes in HER3 expression status between diagnosis and at recurrence. For ovarian and uterine cancers, patients who became HER3-high tended to have a worse prognosis, while those who changed to HER3-low/zero tended to have an improved prognosis (Additional file 7: Figure S3a, b).