Profile of treatment-related complications in women with clinical stage IB-IIB cervical cancer: A nationwide cohort study in Japan

Objective To examine clinico-pathological factors associated with surgical complications and postoperative therapy for clinical stage IB-IIB cervical cancer. Methods This nationwide multicenter retrospective study examined women with clinical stage IB-IIB cervical cancer who underwent radical hysterectomy plus pelvic and/or para-aortic lymphadenectomy between 2008–2009 at 87 institutions of the Japanese Gynecologic Oncology Group (n = 693). Multivariate models were used to identify independent predictors of perioperative grade 3–4 complications and bladder dysfunction. Results The overall intraoperative and postoperative complication rates were 3.3% and 9.8%, respectively. Clinical stage was not associated with perioperative complications (P = 0.15). Radiotherapy-based adjuvant therapy was significantly associated with an increased risk of postoperative complications (radiotherapy alone: adjusted-odds ratio [OR] 3.19, 95% confidence interval [CI] 1.46–6.99, P = 0.004; radiotherapy plus chemotherapy: adjusted-OR 3.26, 95%CI 1.66–6.41, P = 0.001), whereas chemotherapy was not (P = 0.45). Nerve-sparing surgery significantly reduced the risk of postoperative bladder dysfunction (adjusted-OR 0.57, 95%CI 0.37–0.90, P = 0.02) whereas adjuvant chemotherapy increased the risk of bladder dysfunction (adjusted-OR 2.06, 95%CI 1.16–3.67, P = 0.01). Among women receiving adjuvant chemotherapy, nerve-sparing radical hysterectomy significantly reduced the risk of bladder dysfunction (15.0% versus 32.9%, OR 0.31, 95%CI 0.14–0.68, P = 0.004). After propensity score matching, survival outcomes were similar with both types of adjuvant therapy (radiotherapy-based versus chemotherapy, P>0.05). Conclusion Our study highlighted two distinct complication profiles of adjuvant therapy after radical hysterectomy for clinical stage IB-IIB cervical cancer, with radiotherapy increasing grade 3–4 adverse events and chemotherapy increasing bladder dysfunction. In this setting, nerve-sparing surgery may be useful if chemotherapy is being considered for adjuvant therapy.


Study design and patient eligibility
A nationwide retrospective study was conducted at 87 member institutions of the Japanese Gynecologic Oncology Group (JGOG-1070s) [16]. After Institutional Review Board approval (Tokai ethical committee:12R233) was obtained, the institutional cervical cancer database was utilized to examine patients with cervical cancer at each participating institution. In brief, we used the Japanese Gynecologic Oncology Group (JGOG) data, a nationwide study composed by the Committee on Treatment Guidelines for Cervical Cancer in Japan. All data were fully anonymized before access. Eligible women had clinical stage IB-IIB cervical cancer and underwent type III RH with pelvic and/or para-aortic between January 1, 2008 and December 31,2009. The data acquisition period at participating institutions was set from March 1 to August 31, 2012. This study captured consecutive cases of cervical cancer (maximum 10 cases) per in each institution during the study period to reduce selection-bias between institutions and reflects on the reality in clinical practice in Japan. Patients were excluded from this study if their lymph node status was unknown, if the cancer stage was unknown, or if the tumor was stage IA or stages III-IV. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for retrospective observational studies were followed [17].

Clinical information
The following information was obtained from the medical records of eligible patients: (i) demographic data including the age and body mass index (BMI) at diagnosis, (ii) pathological data including the histological subtype, International Federation of Gynecology and Obstetrics (FIGO) cancer stage, tumor size, nodal involvement, and length of the resected vaginal cuff, (iii) treatment information including neoadjuvant therapy, surgical procedures (radical hysterectomy, bilateral salpingo-oophorectomy, para-aortic lymphadenectomy, number of resected lymph nodes, nerve-sparing technique, estimated blood loss, and operating time), and adjuvant therapy (radiotherapy and chemotherapy), (iv) perioperative complications from surgery to the final follow-up selected on the basis of previous studies (ureteral injury, bladder injury, bowel injury, surgical site fistula, urination difficulty, urinary incontinence, lymphocele, lymphedema, lymphorrhea, lower extremity cellulitis, thromboembolism, constipation, diarrhea, intestinal bleeding, and hematuria) [18], and (V) survival outcomes including diseasefree survival (DFS) and cause-specific survival (CSS).

Study definitions
Age and BMI were dichotomized according to previous studies [19]. Histologic subtypes were classed as either squamous or non-squamous. The resected vaginal cuff length was dichotomized at the median value. Complications were graded according to the National Cancer Institute's Common Terminology Criteria for Adverse Events (CTC-AE) version 4.0 [20]. Intraoperative and postoperative complications were defined as adverse events of grade 3 or higher, which occurred the period of a patient's surgical procedure and the period between after surgery and last follow-up. Postoperative bladder dysfunction was defined as difficulty with urination or urinary incontinence requiring prolonged self-catheterization.
RH was either type III or type C hysterectomy, characterized by wide extirpation of parametrial tissue and separation of the posterior leaf of the vesico-uterine ligament [21]. Nervesparing surgery was defined as preservation of the vesical branches arising from the inferior hypogastric plexus while transecting the uterine branches of this plexus [22]. DFS was defined as the interval between surgery and first recurrence or death from cervical cancer, while CSS was the interval between surgery and death from cervical cancer. Patients without these survival events at final follow-up were censored.

Statistical analysis
The primary outcome of this study was the frequency of perioperative adverse events (grade �3) associated with each postoperative treatment modality. Secondary outcomes were the frequency of adverse events for each clinical stage and the interaction of factors contributing to bladder dysfunction. In addition, survival was assessed based on adjuvant therapy types.
Continuous variables were expressed as the mean ± standard deviation (SD) or as the median with interquartile range (IQR). The significance of differences between continuous variables was assessed with the Kruskal-Wallis H test or one-way ANOVA as appropriate. Categorical and ordinal variables were expressed as the number (%), and the significance of differences was examined by the chi-square test.
Binary logistic regression models were used to identify independent risk factors for grade 3-4 perioperative complications or bladder dysfunction. Variables found to be significant on univariate analysis were entered into the initial model. The extent of significance was expressed as the adjusted odds ratio (OR) and its 95% confidence interval (CI).
Survival analysis was performed to examine the influence of adjuvant therapy on DFS and CSS after propensity score matching was performed to adjust for differences of background factors between the groups [23]. The propensity score was determined by multivariate logistic regression analysis, and an automated algorithm was used for one-to-one propensity score matching between the chemotherapy and radiotherapy-based groups. The optimal caliper width for estimating differences was calculated as one fifth of the standardized difference for the logit of the propensity score, resulting in a propensity score difference cut-off value of 0.03 in this study [24]. Standardized difference of �0.10 represented a good balance between the two groups after the matching.
The Kaplan-Meier method was utilized to construct survival curves, and the significance of differences between the curves was assessed by the log-rank test. Multivariate analysis with Cox proportional hazards regression models was used to identify independent prognostic factors for DFS and CSS. All the covariates found to be significant by univariate analysis were entered in the final model, and the extent of significance was expressed as the adjusted hazard ratio (HR) and its 95%CI.
The variance inflation factor (VIF) among the covariates in multivariate analysis was determined, and a VIF �2.0 was defined as indicating multicollinearity. All statistical tests were based upon two-tailed hypotheses, and P<0.05 was considered to indicate statistical significance. The Statistical Package for the Social Sciences (IBM SPSS, version 25.0, Armonk, NY, USA) was used for all analyses.

Factors related to perioperative complications
According to univariate analysis, the factors associated with grade 3-4 intraoperative complications were para-aortic lymphadenectomy, a larger number of resected lymph nodes, and a longer operating time (all P<0.05) ( Table 2). Clinical stage was not associated with an increased risk of grade 3-4 complications (P>0.05). Women who received radiotherapy had the highest incidence of grade 3-4 postoperative complications (14.7% with radiotherapy alone, 16.5% with radiotherapy plus chemotherapy, 7.1% with chemotherapy alone, and 5.3% with no adjuvant therapy; P<0.001). Adjuvant radiotherapy significantly increased the incidence of urinary tract fistula, lymphocele, and ileus (all P<0.05) (S1 Table). Multivariate analysis confirmed that radiotherapy was an independent risk factor for grade 3-4 postoperative complications (adjusted-ORs 3.19-3.26, P<0.05) ( Table 3). On the other hand, chemotherapy was not associated with postoperative complications (P = 0.69).
The interaction between nerve-sparing surgery and chemotherapy was examined (Table 5). According to multivariate analysis, nerve-sparing RH was associated with a significantly lower risk of bladder dysfunction among women receiving postoperative chemotherapy (15.0% versus 32.9%, adjusted-OR 0.31, 95%CI 0.14-0.68, P = 0.004).

Survival outcomes
Survival analysis was performed in the 693 women with cervical cancer. The median follow-up time of censored cases in the whole cohort was 40.0 months (IQR 36.4-44.8) and 69 women died of cervical cancer during the follow-up period. According to multivariate analysis (S2 Table), higher clinical stage, non-squamous histology, non-nerve sparing RH, and lymph node metastasis were independent predictors for decreased survival (all adjusted P<0.05). On the other hand, adjuvant therapy was not an independent predictor of an increased risk of causespecific mortality. However, tumor characteristics were significantly different between the women who received postoperative therapy and those who only underwent surgery. Therefore, we subsequently performed stratified analysis. Survival outcomes were examined in the 427 women who received postoperative therapy. The median follow-up time of censored cases was 40.2 months (IQR 37.9-45.9) for the whole cohort and 62 women died of cervical cancer during the follow-up period. After propensity score matching (S3 Table), demographic factors were similar between the chemotherapy and radiotherapy-based groups (all, SD<0.1). The chemotherapy group and the radiotherapybased group showed similar survival rates (5-year DFS rates, 67.6% versus 63.3%, P = 0.67; and 5-year CSS rates, 79.4% versus 77.0%, P = 0.92 S1A and S1B Fig). On multivariate analysis (S4 Table), higher clinical stage, non-squamous histology, and lymph node metastasis were independent predictors for decreased survival (all adjusted P<0.05). In addition, the influence of nerve-sparing surgery and chemotherapy on survival was examined (S5 Table). According to multivariate analysis, nerve-sparing RH without postoperative chemotherapy was associated with significantly better survival (5-year CSS rates 95.7%, adjusted-OR 0.24, 95%CI 0.10-0.55, P = 0.001).

Discussion
Our study identified two distinct complication profiles related to adjuvant therapy after RH for clinical stage IB-IIB cervical cancer. In brief, we found that radiotherapy was associated with more grade 3-4 adverse events, whereas chemotherapy was associated with a higher risk of bladder dysfunction. Additionally, our results suggested that nerve-sparing surgery reduced bladder dysfunction in women receiving adjuvant chemotherapy.
Randomized clinical trials have demonstrated a survival benefit of adjuvant radiotherapy for high-risk and intermediate-risk early cervical cancer (GOG-92 and GOG-109) [14,25]. A recent JGOG survey showed that chemotherapy was the most common adjuvant therapy for intermediate-risk early cervical cancer, and nearly one third (33.1%) of JGOG-affiliated  institutions treat this disease with adjuvant chemotherapy alone [3]. Other JGOG studies performed in patients with high-risk and intermediate-risk early cervical cancer have demonstrated similar survival with postoperative adjuvant chemotherapy and postoperative radiotherapy [26,27]. While postoperative adjuvant therapy may improve survival in women with early cervical cancer, it is associated with an increased risk of severe adverse events after RH [18,28]. In fact, various adverse events have been reported after type III RH [12][13][14]29]. In particular, radiotherapy is known to be associated with an increased risk of various complications after RH [12,18]. Grade 3-4 adverse events were reported in 6-18% of women with stage IB-IIA cervical cancer who received adjuvant radiotherapy following radical hysterectomy [14] [25]. The present study also demonstrated similar findings, since radiotherapy was associated with the highest postoperative complication rate of nearly 15% after RH with different types of adjuvant therapy, including chemotherapy. It has been reported that performing chemotherapy after RH may increase the risk of intestinal and urologic complications [5]. Postoperative urinary dysfunction after RH may be attributable to damage to the pelvic autonomic nerves, so nervesparing surgery was previously suggested to reduce the risk of urinary complications [22]. Our results support the concept that nerve-sparing RH may reduce postoperative bladder Abbreviation: OR, odds ratio; CI, confidence interval; BMI, body mass index; PAN, para-aortic lymph node; PLN, pelvic lymph node; RT, radiotherapy; and CT, chemotherapy. https://doi.org/10.1371/journal.pone.0210125.t003 Profile of complications in stage IB-IIB cervical cancer Profile of complications in stage IB-IIB cervical cancer dysfunction related to chemotherapy-induced peripheral neuropathy [29,30]. Conversely, women who received postoperative chemotherapy without nerve-sparing RH had a markedly higher incidence of postoperative bladder dysfunction (32.9%).
Since survival was similar with adjuvant radiotherapy or chemotherapy, treatmentrelated complications that can significantly affect the quality-of-life are an important consideration when choosing adjuvant therapy [26], [27]. A recent randomized clinical trial compared postoperative chemotherapy (paclitaxel/cisplatin) with CCRT for high-risk and intermediate-risk early cervical cancer, showing comparable survival in the two treatment arms but better quality-of-life with postoperative chemotherapy [31]. Because these results are favor for chemotherapy as the choice for adjuvant therapy, performing nervesparing RH could be a key to reducing chemotherapy-related bladder dysfunction. In Japan, the JGOG is currently conducting a phase III clinical trial that compares CCRT with chemotherapy after RH for high-risk clinical stage IB-IIB cervical cancer (JGOG-1060) [16], and the trial should provide useful information regarding the efficacy and toxicity of adjuvant chemotherapy in this population.
The strengths of the present study include one of the largest sample sizes reporting perioperative complication to date and sufficient events to perform multivariate analysis. Complication rates are similar to what reported in literature, but the statistics demonstrated in this nationwide large cohort is more valid and informative. In addition, this study was conducted at JGOG-affiliated institutions with accredited quality of care [32]. Limitations include a retrospective design and the possibility of unknown confounders influencing the results of analysis. For example, the reasons for selecting each type of adjuvant therapy were not included in the database. Another weakness is that this study lacks information regarding preoperative bladder function, and thus, the direct association between radical hysterectomy and postoperative bladder function remain unknown. In addition, this study did not evaluate the objective quality of RH or nerve-sparing surgery.
A clinical application of our findings may be in the area of postoperative treatment. This study clearly showed that adjuvant radiotherapy increased the risk of grade 3-4 postoperative complications after type III RH whereas chemotherapy did not, which can be useful information for selecting adjuvant therapy after RH. In addition, adjuvant chemotherapy was associated with an increased risk of bladder dysfunction, while nerve-sparing RH reduced this risk in patients receiving adjuvant chemotherapy. Thus, nerve-sparing surgery would preferably be performed when chemotherapy is to be given postoperatively. Further investigation will be required to confirm these results.  Table. Demographics for propensity score matching. Number (%) per column or median (interquartile range) is shown. Standardized differences are shown before and after propensity score matching. � 6 cases were missing original data and 3 cases were missing after propensity score matching. Abbreviations: SD, standardized difference; and BMI, body mass index. (PDF) S4 Table. Multivariate analysis for survival after propensity score matching. A Cox proportional hazard regression model for multivariate analysis. Significant covariates in the univariate analysis were initially entered the multivariate model. Significant P-values are emboldened. Abbreviations: HR, Hazard ratio; 95%CI, 95% confidence interval; 5-yr (%), 5-year survival; BMI, body mass index; and PAN, para-aortic lymph node. (PDF) S5 Table. Multivariate analysis in survival for women in cervical cancer based on nervesparing surgery and chemotherapy use. Cox proportional hazard regression models for multivariable analysis is shown based on combination patterns of nerve-sparing surgery during radical hysterectomy and postoperative chemotherapy use. Significant covariates in the univariate analysis were initially entered the multivariate model. Significant P-values are emboldened. Abbreviations: 5-yr (%), cumulative risk of 5-year; HR, Hazard ratio; 95%CI, 95% confidence interval; SCC, squamous cell carcinoma; Chemo, adjuvant chemotherapy; and n.a. not available. (PDF)