Objective
On the basis of outcome data from concurrent chemoradiotherapy (CCRT) for locally advanced cervical squamous cell carcinoma, the authors developed a nomogram for predicting survival outcome.
Study Design
Two hundred fifty-one eligible patients with International Federation of Gynecology and Obstetrics stage IIB-IVA squamous cell carcinoma of the uterine cervix who underwent CCRT were included for the construction of the nomogram. Predictor variables included age, serum squamous cell carcinoma antigen, tumor size, parametrium invasion, hydronephrosis, bladder/rectum invasion, and lymph node metastases. Internal validation of the nomogram was performed.
Results
A nomogram for predicting the 5 year overall survival for these patients was constructed on the basis of a Cox regression model from 7 parameters. The concordance index was 0.69.
Conclusion
This nomogram is a predictive tool, upon external validation, that can be used to counsel patients in predicting outcomes. The discriminatory ability of the nomogram indicates that this population should not be considered homogeneous with respect to risk of death.
Cervical cancer is one of the most common female malignancies resulting in cancer-related death. Through screening programs, the incidence and mortality of invasive cervical cancer has been greatly reduced; however, each year, there are still a substantial number of patients who are diagnosed with locally advanced cervical cancer (International Federation of Gynecology and Obstetrics [FIGO] stage IIB-IVA).
Published reports from 5 prospective randomized trials show that concurrent chemoradiotherapy (CCRT) improves local control and increases the survival rate by 30-50% in patients with advanced stage cervical cancer. The consistent findings of these 5 clinical trials have led the United States National Cancer Institute to recommend that patients requiring radiation therapy for the management of cervical cancer should receive concurrent cisplatin-based chemotherapy. Hence, CCRT can be given to a wide range of patients at different stages of cancer, from the relatively good prognosis stage IIB to poor prognosis stage IVA.
Currently the prognosis prediction for advanced cervical cancer is largely based on stage alone, with a 5 year survival rate of approximately 65-69% for stage II, 40-43% for stage III, and 15-20% for stage IV. However, stage alone is unable to provide an integrated perspective in terms of survival; thus, integration of multiple independent risk factors to better predict survival outcome for advanced stage cervical cancer is a critical issue. For example, tumor size is an important risk factor for stage IIB and III cervical cancer, whereas bilateral parametrial involvement is an important risk factor for stage III cervical cancer.
For the subset of patients with a favorable survival, regular follow-up may be a suitable choice; nevertheless, patients with an unfavorable survival, consolidation chemotherapy, or another novel therapeutic modality may be a suitable choice, after undergoing CCRT.
Traditional assessment of risk factors for cervical cancer treats each risk factor as a discrete entity, and even when adjusted by multivariate analysis, this type of assessment cannot provide a panoramic view of a series of appropriate risk factors. By contrast, nomograms are statistically based tools that provide an estimated probability of a specific outcome by combining weighted risk factors to quantify overall risk with a user-friendly interface.
The nomograms have been constructed to predict various clinical endpoints for patients with different types of cancer. Rather than treating each risk factor as an isolated entity, a nomogram takes the weighted specific value of each risk factor into integrated analysis. A nomogram should theoretically be more specific to each individual patient and thus able to predict specific clinical endpoints more accurately.
Generating predictive models and nomograms for overall survival in advanced-stage cervical cancer is difficult, given the heterogeneous nature of the disease. The purpose of the current study was to develop a nomogram for stage IIB-IVA cervical cancer treated by CCRT, for a long-term follow-up cohort of patients in a tertiary referral center.
Materials and Methods
From January 1999 to December 2006, 284 patients with stage IIB-IVA squamous cell carcinoma of the uterine cervix treated with CCRT were included in the analysis. The study was approved by an institutional review board in the index hospital. All patients underwent either punch biopsy or a loop electrosurgical excision procedure conization of the uterine cervix for histological confirmation.
Of the 284 patients, 105 (36.9%) underwent laparoscopic sampling of pelvic/paraaortic lymph nodes. After central pathology review by an experienced pathologist (C.R.L.), clinicopathological factors were prospectively collected for all patients and were stored as an electronic dataset.
Information collected included age at first diagnosis of cervical cancer, FIGO disease stage, tumor grading, baseline GOG (Gynecologic Oncology Group) status (defined as: 0, normal activity; 1, symptomatic, fully ambulatory; 2, symptomatic, in bed less than 50% of the time), baseline serum squamous cell carcinoma antigen (SCC-Ag) level, tumor size (determined as the largest diameter of the primary tumor measured by computed tomography [CT] or magnetic resonance imaging [MRI] imaging), status of parametrial invasion (determined by CT or MRI), status of hydronephrosis (defined as dilatation of the renal pelvis and calyces diagnosed by renal ultrasound, CT scan, or intravenous pyelography), status of urinary bladder/rectum invasion (confirmed by cystoscopic biopsy or sigmoidoscopic biopsy), status of lymph node metastases (confirmed by either pathological verification or defined as nodal size >1cm by CT or MRI).
CCRT was conducted by using weekly (weeks 1-6) infusions of cisplatin (40 mg/m 2 ) or carboplatin (100 mg/m 2 ) if creatinine clearance glomerular filtration rate was less than 60 mL/min. Within 4 hours after completion of chemotherapy, patients started to receive external beam radiation therapy (EBRT) to the whole pelvic area with a 4-field technique (anteroposterior, posteroanterior, and 2 lateral fields) followed 1-2 weeks later by intracavitary brachytherapy (ICRT) with the aim of delivering a total dose of 80-85 Gy to point A (a reference location 2 cm lateral and 2 cm superior to the cervical os) and 50-55 Gy to point B (pelvic wall).
For EBRT, the treatment machines were the linear accelerator 10 MeV or Cobalt units, and for ICRT, a high-dose rate iridium-192 machine was used. Patients with histologically confirmed or suspected (>1 cm) paraaortic lymph node metastases underwent an extended-field radiation technique with a dose of 1.5 Gy per day on days 1-5 of each week, for a total of 30 fractions (45 Gy). Chemotherapy was discontinued if the leukocyte counts dropped below 3000 per cubic millimeter or platelet counts dropped below 100,000 per cubic millimeter.
After completion of concurrent chemoradiation therapy, patients were followed up at 3 month intervals for the first 2 years and at 6 month intervals for another 2 years, and then followed up annually. Of the 284 patients, 65 (22.8%) received consolidation chemotherapy after CCRT at the discretion of the attending physician to extirpate potential micrometastases. The regimen for consolidation included ifosfamide plus cisplatin, bleomycin plus ifosfamide plus cisplatin, or paclitaxel plus cisplatin, with the aim of delivering a total of 4-6 cycles. Overall survival was counted from initiation of chemoradiation to disease-specific death or last follow-up.
Initially, we tried to include adenocarcinoma and adenosquamous carcinoma while building our nomogram but eventually discovered that it would result in a decrease in the concordance index; hence, they were excluded. In addition, these 2 histological types of cervical cancer were not incorporated in our study because several reports have pointed out that they are independent risk factors for advanced stage cervical cancer.
Patients who had incomplete clinicopathological data or a GOG performance score greater than 2 were also excluded from the analysis. The eligible patients formed the basis for constructing the nomogram.
Statistical methods
Descriptive statistics, including mean (or median as appropriate) and range were used to summarize the patient’s characteristics. Overall survival probability was estimated by the Kaplan-Meier method. Multivariate Cox proportional hazards models with stepwise backward variable removal were fit to test the relationship between survival time and identified risk factors. Proportional hazards assumptions were verified systematically for the final model by using the Grambsch-Therneau residual-based test.
The nomogram was constructed as described by Kattan et al. The entered parameters for developing the nomogram included: age, baseline SCC-Ag, tumor size, parametrium invasion, hydronephrosis, bladder/rectum invasion, and lymph node metastases.
Validation of the nomogram was comprised in 2 steps. First, discrimination was measured by generating the concordance index, which is the probability that, given 2 randomly selected patients, the patient who died first had the worst predictive outcome from the nomogram.
Bootstrapping methods were used to obtain a relatively unbiased estimate. Second, calibration was generated by plotting actual survival against the nomogram predicted survival. An ideal plot corresponds to a 45° line, in which the predicted survival probability parallels the actual survival. All analyses were performed using S-plus 2000 Professional software (Statistical Sciences, Seattle, WA).
Results
Two hundred fifty-one eligible patients with stage IIB-IVA squamous cell carcinoma of the uterine cervix who underwent CCRT were included for the construction of the nomogram. Descriptive patient characteristics are shown in Table 1 . Median follow-up for this cohort of patients was 6.3 years. The mean (SD) age of the study population was 48.6 (9.3) years and the majority of patients (71.3%) had a GOG performance score of 1 or less. Overall survival by stage was plotted using the Kaplan-Meier method ( Figure 1 ).
| Variable | Patients (n = 251) | |
|---|---|---|
| Number | % | |
| Age | ||
| Mean ± SD | 48.6 ± 9.3 | |
| Stage | ||
| IIB | 133 | 52.9 |
| IIIA | 19 | 7.6 |
| IIIB | 75 | 29.8 |
| IVA | 24 | 9.5 |
| GOG performance score | ||
| 0-1 | 179 | 71.3 |
| 2 | 72 | 28.7 |
| SCC-Ag | ||
| ≤1.5 | 30 | 11.9 |
| 1.6-5 | 138 | 54.9 |
| 5.1-15 | 61 | 24.3 |
| >15.1 | 22 | 8.7 |
| Tumor grade | ||
| 1 | 32 | 13.2 |
| 2 | 51 | 21.0 |
| 3 | 160 | 65.8 |
| Tumor size | ||
| ≤4 cm | 49 | 19.5 |
| >4 cm | 202 | 80.5 |
| Parametrial involvement | ||
| (–) | 33 | 13.1 |
| (+) | 218 | 86.8 |
| Hydronephrosis | ||
| (–) | 160 | 63.7 |
| (+) | 91 | 36.3 |
| Nodal status | ||
| (–) | 142 | 56.6 |
| (+) pelvic | 71 | 28.3 |
| (+) paraaortic | 38 | 15.1 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
