This retrospective study was performed in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Sun Yat-sen University Cancer Center Institutional Review Board (IRB-number B2020-345-01). Informed patient consent was waived in this observational and noninterventional study, in which 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.researchdata.org.cn), with the approval RDD number as RDDA2021001939.

In this study, a total of 583 patients with FIGO stage IB∼IIA cervical cancer who had received radical resection at Sun Yat-sen University Cancer Center between February 2009 and December 2017 were enrolled as a cohort. Every patient’s postoperative histopathological type was reviewed according to the WHO criteria by two experienced gynecological pathologists. Routine blood examinations were performed before surgery. The study excluded patients who had hematological malignancies, other malignancies, autoimmune diseases, active infection, or inflammatory disease within the month before blood examinations, or who lost contact. The tumor stage of all cases was defined according to the FIGO 2018 cervical cancer criteria [6].

Clinical data, including clinicopathological parameters (such as age at diagnosis, height, weight, histopathological type, tumor grade, maximal tumor size, depth of invasion, preoperative neoadjuvant chemotherapy, postoperative adjuvant therapy, FIGO stage, and HPV infection status), survival time, and preoperative routine blood examination results (including absolute counts of leukocytes, lymphocytes and platelets, the levels of CRP and ALB) were collected retrospectively from the medical records. In addition, we also collected the postoperative adjuvant treatment including radiotherapy and chemotherapy.

Venous blood was drawn from all patients within 1 week before surgery. The CAR, PLR, PNI and body mass index (BMI) were calculated using the following equations: CAR = CRP ( mg / L ) ALB ( g / L ) PLR = platelets   counts ( 10 9 / L ) lymphocytes   counts ( 10 9 / L ) PNI = ALB ( g / L ) + 5 ∗ lymphocytes   counts ( 10 9 / L ) BMI = weight ( kg ) height 2 ² ( m 2 ² ) The   depth   of   invasion = tumor   invasion   in   mm cervical   wall   thickness   in   mm The depth of invasion was evaluated by two experienced gynecological pathologists. The invasion of outer one-third (≥2/3) of the cervical stroma was considered as a risk factor that increases the risk of pelvic recurrence [6].

Receiver operating characteristic (ROC) curves were generated for pre-treatment CAR, PLR, and PNI to determine the cut-off values that yielded optimal sensitivity and specificity. The upper right point of the ROC curve was selected, and the Youden index was calculated according to the sensitivity and specificity of each possible point in the statistical results. Then, the largest point was selected to determine the optimal cut-off value of the hematological index, such that the numerical variable could be transformed into a classified variable to analyze the statistical data.

Every patient accepted high-risk type HPV test before surgery. Cervical cells sample was collected by the Digene Hybrid Capture 2 (HC2) HPV DNA Collection Device (QIAGEN, Germantown, MD, United States), which consisted of a cervical brush and a digene specimen transport medium. The DNA copy number of the high-risk type HPV was calculated. If the copy number was 0–1 pg/ml, it indicated that there was no HPV infection. When the copy number was greater than 1 pg/ml, it indicated a positive HPV infection.

Preoperative neoadjuvant chemotherapy was administered to inhibit some micrometastasis and distant metastasis and to improve the radical curability and safety of surgery. Patients whose tumor could not be removed completely in the beginning were given 2 cycles of cisplatin (70 mg/m²) and paclitaxel (175 mg/m²), every 3 weeks before surgery.

According to the FIGO 2018 and NCCN guidelines, patients with either positive surgical margins or lymph node metastases or parametrial spread, were offered postoperative radiotherapy with chemotherapy. For adenocarcinoma and adenosquamous carcinoma patients with two of four risk factors (deep stromal invasion, tumor size more than 3 cm, lymphovascular invasion, adenocarcinoma), and for squamous cell carcinoma patients with two of three risk factors (deep stromal invasion, tumor size more than 4 cm, lymphovascular invasion), postoperative radiotherapy was required. Among other patients, those with poorly differentiated grade and tumor size more than 2 cm, or highly maligant pathological type such as clear cell adenocarcinoma, neuroendocrine carcinoma, were given chemotherapy without radiotherapy. All other patients following radical hysterectomy did not need any adjuvant therapy [6, 26].

After the initial treatment, follow-up assessments were performed once every 3 months for the first 2 years, once every 6 months for the next 3 years, and then once every year after 5 years. The main examinations included routine blood tests, tumor markers measurement, gynecological examinations, chest radiography, and computed tomography (CT) or magnetic resonance imaging (MRI) scans. In cases with abnormal findings and suspected tumor recurrence, lesion biopsy was performed to determine whether recurrence had occurred. As study endpoints, OS was defined as duration from the date of initial surgery to the date of death for any causes or the last follow-up (December 31st, 2020), and PFS was defined as the time interval from the date of surgery to progression or relapse.

Comparison analyses of inflammatory indexes and clinicopathological variables between HPV-positive and HPV-negative groups were conducted using t-test, Pearson’s chi-squared (χ²) test and Fisher’s exact test. Comparison analyses of clinicopathological variables between high- and low-inflammatory index groups were performed as well.

The survival curves were plotted using the Kaplan–Meier method, and log-rank tests were carried out to assess survival differences between groups. Significant prognostic variables in the univariate analysis were selected for the multivariate Cox Regression model analysis to identify independent prognostic variables using the forward stepwise method. Analyses were performed using SPSS 23.0 software (IBM Corporation, Armonk, NY, United States) and GraphPad Prism 8 software (GraphPad Software, Inc., San Diego, CA, United States). A two sided p < 0.05 indicated that the difference was considered statistically significant.

In total, 583 patients with stage IB∼IIA cervical cancer were recruited in this study, of which 501 (85.93%) cases were HPV-positive and 82 (14.07%) cases were HPV-negative. The associations of characteristics between two groups were shown in Table 1. All patients had no lymph node metastasis. The median age at the time of prognosis was 48.88 ± 9.174 years, and the median follow-up time was 68.04 ± 26.32 months. The HPV-positive group had a higher proportion of squamous cell carcinoma type compared to HPV-negative group (χ² = 35.962, p < 0.0001). The proportions of other characteristics were relatively balanced. During the follow-up period, a total of 42 patients experienced recurrence or progression, and 31 of them died.

In this study, 109 HPV-positive cases and 24 HPV-negative cases received neoadjuvant chemotherapy before surgery. Patients with risk factors were required to accept radiotherapy or chemotherapy after surgery. In HPV-positive group, there were 151 cases without postoperative adjuvant therapy, 96 cases with chemotherapy, 111 cases with radiotherapy and 143 cases with chemotherapy and radiotherapy. In HPV-negative group, there were 28 cases without postoperative adjuvant therapy, 17 cases with chemotherapy, 12 cases with radiotherapy and 25 cases with chemotherapy and radiotherapy. We compared the prognosis of different subgroups and found that there was no influence on prognosis of cervical cancer patients no matter what kind of postoperative adjuvant therapy they received (Supplementary Figure S1).

The cut-off values of CAR, PNI, and PLR were determined using ROC curves (Supplementary Figure S2). The analyses identified CAR ≤ 0.0396 (area under the curve (AUC): 0.619, 66.7% sensitivity, 61.9% specificity), PLR ≤ 163.41 (AUC: 0.509, 33.3% sensitivity, 78.9% specificity), PNI ≤ 50.15 (AUC: 0.520, 66.7% sensitivity, 21.3% specificity), as the most accurate cut-off values for patients with cervical cancer.

Patients with different HPV infection statuses were divided into two groups according to the cut-off values of CAR, PNI, or PLR. The associations of clinical characteristics between CAR^hi and CAR^lo, PLR^hi and PLR^lo, PNI^hi and PNI^lo groups were analyzed using t-test, chi-squared (χ²) test and Fisher’s exact test in HPV-positive group and HPV-negative group, respectively.

We found that although there were no differences in age, FIGO stage, and pathological type among patients with HPV-positive cervical cancer. A higher CAR level (>0.0396) correlated significantly with larger tumor size (χ² = 8.360, p = 0.004), higher receipt rate of neoadjuvant chemotherapy (χ² = 6.350, p = 0.012) and nonstandard BMI (χ² = 16.962, p < 0.0001) (Table 2). A higher level of PLR (>163.41) was associated with larger tumor size (χ² = 4.086, p = 0.043) and lower PNI (χ² = 49.013, p < 0.0001) (Table 3). While a higher level of PNI (>50.15) was associated with smaller tumor size (χ² = 6.139, p = 0.013), lower receipt rate of neoadjuvant chemotherapy (χ² = 4.190, p = 0.041) and lower PLR (χ² = 49.013, p < 0.0001) (Table 4).

For HPV-negative group, the results showed that a higher CAR level was related to nonstandard BMI (χ² = 6.786, p = 0.009) (Table 5), and there was no difference of clinicopathological variables between PLR^hi and PLR^lo groups (Table 6) as well as PNI^hi and PNI^lo groups (Table 7).

Kaplan–Meier analyses showed that there was no difference of prognosis between HPV-positive and HPV-negative groups (Supplementary Figure S3). Among 501 patients with HPV-positive cervical cancer, CAR^hi (p < 0.0001; p = 0.0009), PLR^hi (p = 0.0021; p = 0.0108) had worse OS and PFS (Figures 1A–D), while PNI^hi (p = 0.0013; p = 0.0403) had favorable OS and PFS (Figures 1E,F). For the HPV-negative group, none of the three indexes were related to prognosis (all p > 0.05) (Supplementary Figure S4).

For HPV-positive group, the multivariate analysis showed that the depth of invasion (HR: 3.651, 95% CI: 1.464–9.103, p = 0.005), the levels of PNI (HR: 0.341, 95% CI: 0.156–0.745, p = 0.007) and CAR (HR: 5.201, 95% CI: 2.080–13.004, p < 0.0001) were independent factors for OS (Table 8). In addition, the depth of invasion (HR: 2.478, 95% CI: 1.218–5.043, p = 0.012), the levels of PLR (HR: 1.991, 95% CI: 1.018–3.894, p = 0.044) and CAR (HR: 2.769, 95% CI: 1.406–5.455, p = 0.003) were independent predictors of PFS (Table 9). For the HPV-negative group, only the depth of invasion (HR: 9.192, 95% CI: 1.016–83.173, p = 0.048) was an independent prognostic factor of OS (Table 10), while none of clinicopathological parameters were independent prognostic indicators of PFS (Table 11).

All the above data suggested that for patients with HPV-positive FIGO stage IB∼IIA cervical cancer, the depth of invasion, the levels of PNI and CAR are independent predictors of OS, and the depth of invasion, the levels of PLR and CAR are independent predictors of poor PFS. For patients with HPV-negative cervical cancer, none of CAR, PNI and PLR is independent factor of prognosis.