Impact of the initial site of metastases on post-recurrence survival for neuroendocrine cervical cancer

Objective To summarize the risk factors and emphasize the prognostic importance of the site of recurrent neuroendocrine cervical cancer (NECC). Methods We enrolled 88 patients who developed recurrence after radical surgery for pathological stage I–IVa primary NECC between January 2003 and 30 December 2020 and classified these cases into 7 groups based on the initial recurrence. The risk factors for post-recurrence survival (PRS) were analyzed by Kaplan–Meier and Cox regression methods. Results Among 88 NECC patients, nearly all patients (95.50%) experienced progression within 3 years. The time to progression was significantly longer in patients with lung recurrence than in patients without lung recurrence (p = 0.008). After the first recurrence, the median follow-up was 11.1 months (range 2.37–65.50 months), and the 5-year PRS was only 20.6%. The depth of invasion in the primary surgery, number of recurrent sites, abdominal organ recurrence were correlated with PRS by univariate analysis. Multivariate analyses revealed that the number of recurrent sites (p = 0.025) and abdominal organ recurrence (p = 0.031) were independent prognostic factors. Notably, the combination of immune checkpoint inhibitors and chemotherapy, with or without surgery, showed a 43.8% objective response rate in recurrent NECC. Conclusion Patients with abdominal organ recurrence need more sophisticated therapy. The combination of immune therapy and chemotherapy might be an opportunity for recurrent NECC. Supplementary Information The online version contains supplementary material available at 10.1186/s12885-022-09737-4.


Introduction
Neuroendocrine cervical cancer (NECC) is rare, accounting for less than 5% of all cervical tumors [1,2]. According to the 2014 World Health Organization (WHO) classification of tumors of the female reproductive organs, NECC has two groups: carcinoid tumors and atypical carcinoid tumors are referred to as lowgrade NECC, and small cell neuroendocrine cervical cancer (SCNECC) and large cell neuroendocrine cervical cancer (LCNECC) are high-grade NECC [3]. The 5-year survival rate of NECC is 37% for the early stages and 9% for the advanced stages [4]. More than 50% of NECC patients relapse within 5 years despite systemic therapies [5,6], which is a much higher rate than that of conventional cervical tumors (20-40%) [7]. Given the Open Access † Baoyue Pan and Ting Wan contributed equally to this work. *Correspondence: zhengmin@sysucc.org.cn 1 State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Sun Yat-Sen University Cancer Center, 651 Dongfeng Road East, Guangzhou 510060, People's Republic of China Full list of author information is available at the end of the article aggressive nature of the NECC, it is important to identity the recurrence pattern and establish a risk model.
However, most studies on the recurrence pattern of NECC are case reports, and the patients collected in these studies are relatively limited to predicting prognosis [8][9][10]. In addition, few studies have reported the response to immune checkpoint inhibitors (ICI) therapy in NECC.
In conventional recurrent cervical tumors, lung metastasis indicates favorable clinical outcomes [11]. The number of distant metastatic sites is one of the independent factors for metastatic pancreatic neuroendocrine carcinoma [12]. With this background, we speculate that the prognosis of recurrent NECC is related to recurrent sites. We summarized the recurrent patterns of NECC in our center and discussed the prognostic factors in this study.

The selection of patients
We reviewed 126 patients who were diagnosed with recurrent NECC at Sun Yat-sen University Cancer Center from 1 January 2003 to 30 December 2020. Among them, 19 patients were followed up for less than 60 days after the first recurrence, 8 patients were at stage IVb according to the International Federation of Gynecology and Obstetrics (FIGO 2018) criteria, and 11 patients did not undergo radical hysterectomy. These 38 patients were excluded from the study. Finally, 88 patients with FIGO stages I-IVa were included in the study.
This retrospective study was performed in accordance with the Declaration of Helsinki and approved to waive informed patient consent by the institutional review board of Sun Yat-sen University Cancer Center (approval number: B2020-330-1) due to the observational and noninterventional study, and the patient's data were kept under strict control. The authenticity of this article has been validated by uploading the key raw data onto the Research Data Deposit public platform (www. resea rchda ta. org. cn), with the approval RDD number as RDDA2021001979.

Baseline data collection
Demographic and clinicopathological data were collected from hospital records, including age, high-risk HPV human papillomavirus (HPV) test, FIGO stage, primary tumor size, pathology, and treatment when the patients were first diagnosed with NECC. When the tumor relapsed, data related to the pattern of recurrence and the follow-up treatment were collected.

Diagnosis of recurrent NECC
The diagnosis of NECC was dependent on immunohistochemical analysis of the primary tumor or metastasis, which was demonstrated using several markers, including chromogranin A, synaptophysin, and CD56. Sometimes, neuron-specific enolase (NSE) was also used. Some patients were referred to as NECC not otherwise specified (NECCNOS) if they did not have a typical morphological classification but exhibited neuroendocrine markers. All available pathological slides were reviewed together by gynecological pathologists at Sun Yat-sen University Cancer Center.
To define recurrence, imaging examinations, such as F-fluoro-2-deoxy-d-glucose (FDG) positron emission tomography and CT (FDG-PET/CT), or contrastenchanced computed tomography (CT), or magnetic resonance imaging (MRI), were employed as necessary investigations. A 20% increase in lesion size on a follow-up imaging scan within 3 months was also defined as recurrence when disease progressed in the primary treatment.

Survival
Time to recurrence (TTR) was determined based on the time from the date of the diagnosis of NECC to the first recurrence by imaging evidence or histology. Post recurrence survival (PRS) was defined as the time from the date of the first diagnosis of recurrence to the follow-up deadline or the date of death.

Immune checkpoint inhibitors (ICI) therapy
The ICI therapy was employed in 22 patients. 16 patients of them received at least 3 cycles and underwent imaging examinations to assess response. 4 patients underwent detection of ICI biomarkers, included PD-1 (1/16), PD-L1 staining (2/16) and tumor mutation burden (1/16). The remained 14 patients all progressed after anti-angiogenesis therapy or several cycles of second-line chemotherapies or local radiotherapy. Because multimodality therapy showed no benefit for patients, doctors introduced the ICI therapy and fully informed the patients of the tradeoffs.

Statistical analysis
For comparisons of groups, the χ2-test and Fisher's exact test were used where appropriate. All survival curves were plotted using the Kaplan-Meier method, and logrank tests were carried out to assess survival differences between groups. Univariate and multivariable forward stepwise Cox regression models were used for OS analysis. A difference of 0.05 was considered significant. SPSS 25.0 (IBM Corp., Armonk NY, USA) was used for the statistical calculations.

Baseline characteristics
The median age of the 88 patients who experienced recurrence was 46 years. Among the 43 patients who underwent HPV-based screening, 93.0% had HPV infection. According to the 2008 FIGO staging system, 50, 22, 0, and 2 patients had stage I, II, III and IVa disease, respectively. We restaged the patients using the 2018 FIGO system, which showed that 42, 9, 27, and 2 patients had stage I, II, III, and IVa disease, respectively. SCNECC was the most common pathological subtype (80.7%), and LCNECC and NECCNOS accounted for only 19.3% of all cases of recurrence in patients. Unexpectedly, nearly a quarter of patients (23/88) had mixed NECC, and mixed adenocarcinoma (AdC) was much more prevalent than mixed squamous carcinoma (SqC). Among all neuroendocrine differentiation markers, synaptophysin, chromogranin A, and CD56 were expressed in 94.0%, 80%, and 90.5% of patients, respectively. In addition, immunohistochemistry showed that 82.7% of 52 patients tested were positive for NSE.
All patients underwent radical surgery, and 29 patients (33.0%) showed pelvic lymph node metastasis (PLNM). In the primary treatment, all patients underwent chemotherapy and approximately 81.8% of them received at least four cycles. In addition, 60.2% patients received adjuvant radiation therapy. Other clinicopathological variables are listed in Table 1.

Distribution of recurrent sites
Hematogenous metastases, especially lung and abdominal organ recurrence, were the most common recurrent sites. There were 34 (38.6%) patients with lung recurrence, 33 (37.5%) patients with abdominal organ recurrence, 24 (27.3%) patients with pelvic organ recurrence, 11 (12.5%) patients with bone recurrence, 10 (11.4%) patients with cervicothoracic lymph node recurrence, 7 (8.0%) patients with brain recurrence and 7 (8.0%) patients with vaginal vault recurrence. Among the patients with abdominal organ recurrence, there were 25 patients with liver metastasis, 5 patients with abdominal lymph node metastasis, 2 patients with adrenal gland metastasis, 2 patients with pancreas metastasis, 1 patient with splenic metastasis and 1 patient with kidney metastasis. Approximately half of the patients had metastases in more than one organ at first recurrence. Detailed distributions of metastatic sites are shown in Table 2.

The impact of site-specific metastases on TTR and PRS
Nearly all patients (95.50%) experienced postoperative progression within 3 years, and 99.86% of patients showed recurrence within 5 years. Only one patient showed recurrence in the 70th month. The TTR of patients with lung recurrence was better than that of patients without lung recurrence (16.1 months vs 11.2 months, p = 0.008) (Fig. 1A) Other recurrent sites have no effect on the TTR (Fig. 1 B-G).
After the first recurrence, the median follow-up was 11.1 months (range 2.37-65.50 months), and the 5-year PRS was only 20.6%. The PRS of patients with abdominal or pelvic organ recurrence was worse than that of the remaining patients without abdominal or pelvic organ recurrence (12.2 vs 26.6 months, p = 0.010 and 18.0 vs 15.5 months, p = 0.044, respectively; Fig. 2A-B). The 3-year PRS was 10.4 and 66.5% among patients with or without abdominal recurrence, respectively. Other recurrent sites have no effect on the PRS (Fig. 2C-G).
We included the depth of invasion at the time of primary treatment and four recurrence-related factors (abdominal organs recurrence, pelvic organs recurrence, number of recurrent sites and ICI therapy) in the multivariate analysis. However, two or more recurrent sites and abdominal organ recurrence were finally correlated with worse outcome (Table 3).
Through further comparing patient characteristics with and without abdominal organ recurrence, we found that elderly patients (p = 0.022) and lung recurrence were observed less frequently in patients with abdominal organ recurrence than in those without abdominal organ recurrence (Table 4).

Treatment of disease recurrence
We collected the detailed treatment of 75 patients after recurrence. Among these patients, 63 underwent chemotherapy, 26 received metastasis-directed radiation, and 20 received palliative surgery, transcatheter arterial chemoembolization (TACE), or radiofrequency ablation (RFA). However, antiangiogenic therapy (16/84) failed to improve survival until the follow-up deadline. Among the 22 patients who received ICI therapy, 16 patients received at least 3 cycles, and the tumor response was assessed by imaging (Table 5). Based on the final tumor response, the objective response rate was 43.8% (7/16). However, among the rest 53 patients who received no PD-1 inhibitors, 41 patients were assessed by imaging and the objective response rate was only 22.0% (9/41).
The time to further progression or the end of followup after PD-1 treatment ranged from 3-14 months. The best tumor response to PD-1 inhibitors was as follows: progressive disease (4/16), stable disease (4/16), partial response (3/16) and complete response (5/16). All patients above received chemotherapy and three of them underwent surgery. The lung and vaginal vault were the most common initial recurrent sites of these patients who benefit from PD-1 treatment.

Discussion
This study categorized the initial metastatic organs into 7 initial recurrence sites and demonstrated that the number of recurrent sites and abdominal organ recurrence were independent poor prognostic factors of PRS. However, other initial recurrence sites did not have any impact on the PRS.
Interestingly, our results showed that nearly onethird of patients were intermixed with AdC. Most cases of NECC have HPV infection (especially HPV 18), and some of them can be asymptomatic because of endophytic growth, which is common in cervical AdC [13]. Other neuroendocrine tumors are more similar to AdC than SqC. For example, in the lung and prostate, transdifferentiations from an AdC to neuroendocrine tumor occur in response to targeted therapy [14,15]. However, in cervical cancer, the development of a neuroendocrine carcinoma from a small human papillomavirus-associated cervical adenocarcinoma has been reported lately [16]. Defining the molecular mechanisms of neuroendocrine transformation in cervical cancer remains a question.
Many studies have described the possible risk factors for primary NECC. FIGO stage has been proven to be a recognized factor [17]. In addition, tumor size, lymph   node metastasis, chemotherapy cycles and other factors have different effects on survival [18]. However, the risk factors that affect prognosis after recurrence have not been investigated. In the present study, we conducted univariate and multivariate analyses, which indicated that the number of recurrent sites and abdominal organ recurrence were the main factors associated with PRS.
The common distant recurrent patterns in NECC are lung, abdominal organs, bone, and brain, which is the same as that of ordinary cervical cancer with hematogenous dissemination [19] and other sites of neuroendocrine tumors [20]. However, we found that lung metastasis had the highest rate among all recurrences in NECC, which is different from ordinary cervical cancer, whose pelvic relapse and distant lymphatic dissemination containing para-aortic lymph nodes or SCLN is relatively common [21,22]. In particular, we revealed that lung metastasis in NECC occurred in the late postoperative period but had no effect on PRS. Only abdominal organ recurrence and pelvic organ recurrence were independent poor prognostic factors of PRS.
The most common primary treatment for NECC is radical surgery combined with chemotherapy in the early stage [23]. Chemotherapy regimens containing etoposide and platinum (EP) are recommended [24]. However, the treatment of recurrent disease is individualized and difficult, even in all cervical cancers [25,26]. Most patients underwent chemotherapy containing platinum with etoposide or paclitaxel after recurrence. Some patients received palliative radiation and surgery. However, in the past decade, targeted therapy, such as anti-angiogenesis and immune therapy, has shown potential for treating resistant and recurrent cervical cancer [27][28][29]. Combinations comprising bevacizumab have been proven to improve the progression-free survival of recurrent SCNECC [30]. Pembrolizumab has also been used in recurrent neuroendocrine carcinoma of the lower genital tract, but when used alone, it shows minimal activity  [31]. Only two case reports describe positive responses to nivolumab in NECC [10,32]. Many patients with NECC received the latest therapy in this study. Regretfully, we obtained no significant results in multivariate analysis, likely due to the limited number of patients and short follow-up. In this study, three patients had PD-1/PD-L1 staining positively and one patient had high tumor burden, they all had disease controlled and three of them showed response to ICI, indicating the importance of biomarker detection [33]. Notably, lung and vaginal vault were slightly positively correlated with the response of PD-1 (p = 0.06) (Supplementary Table 1). In addition, a durable effect of PD-1 inhibitors was observed in two patients, similar to the effect in other tumors [34]. Both of them underwent biomarker detection of ICI therapy showed potential response to it. In addition, one of them had lung recurrence, which was more sensitive to PD-1 inhibitors than liver metastasis [35]. The other patient had the LCNECC histology, whose prognosis is better than that of SCNECC [8]. This retrospective study had inevitable limitations. Firstly, this was a single-center and small sample size study. In addition, selection biases also affected the prognosis of patients. Last but not least, only several patients had achieved genetic testing and not every patient underwent biomarker detection before ICI therapy. However, we collected as many patients as we could to focus on the pattern of recurrence and survival of patients with NECC. Our findings support the latest reports about targeted therapy and immune therapy and provide further insight into ICI therapy. Above all, we think that the pathogenesis and clinical manifestation of NECC are similar to those of cervical cancer. However, because of its special histology, the nature of this disease is more aggressive.

Conclusion
Our results revealed that the number of recurrent sites and abdominal organ recurrence were significant prognostic factors in recurrent NECC. The combination of ICI and chemotherapy might be an opportunity for recurrent NECC on the basis of ICI biomarkers. International multicenter studies for recurrent NECC on various combinations of active chemotherapeutic agents, target therapies and immune therapy were warranted to improve the prognosis of recurrent NECC.