This study sought to determine if treatment time impacts pelvic failure (PF), distant failure (DF), or disease-specific mortality (DSM) in patients undergoing concurrent chemoradiotherapy (CCRT).
This study sought to determine if treatment time impacts pelvic failure (PF), distant failure (DF), or disease-specific mortality (DSM) in patients undergoing concurrent chemoradiotherapy (CCRT).
A retrospective review was performed of 113 consecutive eligible patients with stage IB2 to IIIB cervical cancer. All patients received whole-pelvis radiation with concurrent chemotherapy and consolidative intracavitary brachytherapy (BT) to the cervix, followed by an external beam parametrial boost when appropriate. The effect of treatment time on PF, DF, and DSM was examined with univariate and multivariate analyses. Characteristics of patients with and without treatment prolongation were compared to explore reasons for treatment prolongation.
The median time to completion of BT was 60 days, and the median time to complete all RT was 68 days. The 3-year cumulative incidence of PF, DF, and DSM were 18%, 23%, and 26%, respectively. On multivariate analysis, time to completion of BT >56 days was associated with increased PF (hazard ratio, 3.8; 95% confidence interval, 1.2-16; P = .02). The 3-year PF for >56 days versus ≤56 days was 26% versus 9% (P = .04). Treatment time was not associated with DF or DSM. Treatment prolongation was found to be associated with delay in starting BT and higher incidence of acute grade 3/4 toxicities.
In the setting of CCRT, treatment time >56 days is detrimental to pelvic control but is not associated with an increase in DF or DSM. To maximize pelvic control, we recommend completing BT in 8 weeks or less. Cancer 2013. © 2012 American Cancer Society.
Historically, locally advanced cervical carcinomas were treated with radical radiation therapy (RT) alone using a combination of external beam RT to the whole pelvis and a brachytherapy (BT) boost to the cervix. In 1999, the treatment paradigm shifted to concurrent chemoradiotherapy (CCRT) after the publication of 5 randomized trials that demonstrated a survival advantage with the addition of cisplatin-based chemotherapy to RT in the adjuvant and definitive setting.1-5 A number of patient and tumor characteristics have been found to be prognostic in the setting of RT alone or CCRT. For patients treated with RT alone, the detrimental effect of RT prolongation is well established. Total RT time beyond 7 to 9 weeks results in increased pelvic failures (PFs), decreased cause-specific survival, and decreased overall survival (OS).6-9 On the basis of these data, current Radiation Therapy Oncology Group/Gynecologic Oncology Group (RTOG/GOG) protocols on CCRT for cervical cancer recommend that all RT be completed within 56 days (8 weeks).10, 11 Although this recommendation is based on outcomes for single-modality RT, the impact of treatment time on tumor control and survival has not been well studied in the era of CCRT, and the optimal treatment time remains undefined. Evidence-based recommendations are lacking for coordination of CCRT, BT, and parametrial boost (PMB) and relationship between treatment time and patient outcomes. This is especially relevant because there may be many competing pressures on the part of practitioners and patients to coordinate the timing of chemotherapy administration with radiation.
The biologic rationale for compromised outcomes with RT prolongation has been studied, and it is hypothesized that treatment prolongation allows increased repopulation of cancer cells, resulting in reduced local control rates and decreased treatment efficacy.7, 8 One of the main mechanisms by which chemotherapy could improve local tumor control when given concurrently with RT is to reduce accelerated repopulation of tumor cells.12 Clinical data and mathematical models also indicate that concurrent cytotoxic chemotherapy increases the biological effective dose of RT.13, 14 Therefore, it is plausible that concurrent chemotherapy might mitigate the importance of short overall treatment time in cancer of the cervix.
The primary objective of this study was to perform a retrospective analysis of patients with stage IB2-IIIB cervical cancer who were treated definitively with CCRT and BT to determine if treatment time affected PF, distant failure (DF), or disease-specific mortality (DSM). The secondary objective was to report OS and disease-free survival (DFS) and to analyze reasons for treatment prolongation in this patient population.
We identified 113 consecutively treated eligible patients with FIGO (International Federation of Gynecology and Obstetrics) stage IB2 to IIIB cervical cancer from 1997 to 2009 at the University of Chicago Medical Center and the University of Illinois at Chicago. All data were collected after institutional review board approval, with a waiver of consent. Eligibility criteria included definitive treatment with CCRT and consolidative BT without hysterectomy, completion of the intended treatment, and no para-aortic adenopathy or distant metastasis at the time of diagnosis. Tumor stage was assessed according to the 2009 FIGO staging systems. The staging workup included examination under anesthesia, chest imaging, laboratory analysis, and computed tomography to evaluate the primary tumor extent and lymph node (LN) status. Pelvic LN status was determined by staging diagnostic computed tomography (CT) scans read by dedicated radiologists. LNs were considered positive if the diameter of the short axis measured >1 cm on axial slices. Magnetic resonance imaging and positron emission tomography were not used routinely for staging.
All patients underwent CT simulation for planning. Patients received a median of 45 Gy (range, 40-50.4 Gy) to the whole pelvis (WP) with either a three-dimensional conformal technique (40%) or intensity-modulated RT (IMRT; 60%) as described,15 in 1.8-2 Gy fractions up to 5 times a week. Extended field RT to prophylactically treat the para-aortic LNs was used in only 1 patient. BT boost was delivered using either low-dose rate (LDR) technique with 137Cs (N = 107, 95%) or high-dose rate (HDR) technique with 192Ir (N = 6, 5%). The median LDR dose was 40 Gy (range, 33-47.2 Gy) to point A, which represents paracervical tissue, delivered in either 1 (N = 91, 85%) or 2 (N = 16, 15%) fractions. The median HDR dose was 30 Gy (range, 27.5-30 Gy) to point A in 5 fractions. The median cumulative dose to point A was 85 Gy. After the completion of whole pelvic RT and BT, 66% of patients received a PMB to point B, which represents the lateral pelvic structures such as the obturator nodes, by the use of anterior–posterior/posterior–anterior fields in 1.8- to 2.0-Gy daily fractions. After taking into account the dose contribution from BT, the cumulative median dose to point B was 61 Gy (range, 55-67 Gy). When clinically indicated, PMB was delivered after BT, because the dose prescribed to point B was dependent on the dose already received during BT. It was not possible to accurately predict the dose contribution from BT beforehand, given the variability in individual tumor and implant geometry. BT time was defined from the first day of WP RT to the day the total prescription dose was delivered to point A, which usually corresponded to the completion date of the BT. Total RT time was defined from the first day of WP RT to the last day of RT, which corresponded to BT time plus time to complete PMB.
All patients received weekly intravenous chemotherapy for 5 or 6 cycles, concurrent with pelvic RT. The majority of patients (N = 101, 89%) were treated with cisplatin-based chemotherapy. The regimens included cisplatin 40 mg/m2/week (n = 95), cisplatin 40 mg/m2/week and vinorelbine 15 to 20 mg/m2/week (n = 5), or cisplatin 40 mg/m2/week and irinotecan 0.5 mg/m2/week (n = 1). Non-cisplatin regimens (N = 11, 11%) included vinorelbine at 20 to 25 mg/m2/week (n = 6) or vinorelbine at 20 mg/m2/week and paclitaxel at 20 mg/m2/week (n = 6). Patients received non-cisplatin regimens due to renal dysfunction or enrollment onto institutional phase 1 protocols.16, 17
Patients were monitored during treatment for acute toxicities (up to 30 days after completion of all RT). Toxicities were graded based on RTOG criteria. The highest grade for gastrointestinal, genitourinary, and hematologic toxicities were recorded for all patients (N = 113).
Patients were followed every 3 months for the first 2 years, every 6 months for years 3 to 5, and then annually. The workup during the follow-up period included pelvic examination, Papanicolaou tests, biopsy of accessible sites, and imaging studies.
Death was attributed to cervical cancer when death occurred in the setting of widely metastatic disease and/or locally recurrent disease with obstructive symptoms.
Outcome event times were measured from the completion of RT. OS was defined as the time to death from any cause. DSM was defined as deaths from cervical cancer. PF and DF were defined as the time to the first radiographic and/or pathologic evidence of disease recurrence within the pelvis (PF) or outside of the pelvis (DF). Abnormal Papanicolaou tests were not designated as PF, unless verified by cervical biopsy. DFS was defined as the time to the first evidence of PF, DF, or death from any cause. Patients not having a DFS event were censored at the last known medical encounter. Statistical analysis was performed using JMP Software (version 7.0; SAS Institute, Cary, NC). Both BT and total RT time were examined for association with clinical outcome. Fifty-six days (8 weeks) was used to stratify treatment time, because this was the recommended time to complete RT in contemporary RTOG/GOG protocols.10, 11 The Kaplan-Meier method was used to analyze OS, DFS, PF, DF, and DSM. Univariate analysis was performed using the log-rank test. Variables found to have a P value ≤.1 on univariate analysis were entered into multivariable analysis, which was performed using the Cox proportional hazards model. Baseline characteristics were compared using chi-square test and 2-sided t test. RT durations were compared using the Wilcoxon-Mann-Whitney test. A P value of ≤.05 was considered statistically significant.
Tumor and patient characteristics of the 113 patients treated with CCRT are shown in Table 1. The median age at diagnosis was 49 years. The majority of the patients had FIGO stage II (57%) and III (24%) disease with squamous cell histology (93%). Seventeen percent had positive pelvic LNs based on CT scan criteria. Table 2 lists treatment time in days. The median time to complete WP RT was 35 days (range, 30-66 days). The median time to complete BT and all RT were 60 days (range, 39-113 days) and 68 days (range, 45-123 days), respectively. There was a median of 17 days (range, 13-26 days) between the last day of WP RT and the start of the first BT fraction (WP RT-BT interval).
|Characteristics (N = 113)||Value|
|Median (range)||49 (20-92)|
|Age group, y|
|Pelvic LN statusa|
|Squamous cell carcinoma||105 (93%)|
|Median tumor size, cm||5.5 cm|
|Median baseline hemoglobin, g/dL||11.7|
|Median (Range)||IQR||Mean (SD)||No. ≤56 days (%)|
|WP RT time||35 (30-66)||32, 37||36 (5)||110 (97)|
|WP RT-BT time||18 (–30-60)a||13, 26||20 (13)||–|
|BT time||60 (39-113)||52, 68||62 (14)||49 (43)|
|Total RT time||68 (45-123)||58, 80||70 (16)||22 (21)|
The median follow-up time for all patients was 26 months (range, 3-136 months; intraquartile range [IQR] = 14, 51 months). The median time to a recurrence was 7 months (range, 1-22 months; IQR = 4, 16 months). The median follow-up for patients without recurrence was 40 months (range, 3-136 months; IQR = 22, 59 months). PF occurred in 12 patients (11%), DF occurred in 20 patients (18%), and both PF and DF occurred in 4 patients (4%). A total of 27 patients (24%) died from their disease: 16 had DF, 8 had PF, and 3 had both DF and PF. The 3-year OS for all patients was 66% (95% confidence interval [CI], 9%), and for stage I, II, and III patients were 60%, 71%, and 62%, respectively (P = .2). The 3-year DFS for all patients was 58% (95% CI, 9%), and for patients with stage I, II, and III disease were 57%, 67%, and 43%, respectively (P = .01). The 3-year cumulative incidence of PF as first events for all patients was 18% (95% CI, 7%), and for patients with stage I, II, and III disease was 16%, 17%, and 20%, respectively (P = .2). The sites of PF were cervix/parametrium in 12 patients, pelvic LN in 2 patients, and vagina in 2 patients. The 3-year cumulative incidence of DF as first events for all patients was 23% (95% CI, 8%), and for patients with stage I, II, and III disease were 20%, 16%, and 46%, respectively (P < .05). Half of the patients with DF failed at multiple distant sites rather than a single site at the time of failure. The 3-year cumulative DSM was 26% (95% CI, 8%), and for patients with stage I, II, and III disease were 27%, 22%, and 36%, respectively (P = .1).
Patient/tumor characteristics and treatment times were examined on univariate analysis for association with PF, DF, and DSM (Table 3). Time to completion of BT (BT time) was found to be significantly associated with PF, but not with DF or DSM. OS and DSF were also not affected by BT time (data not shown). The 3-year cumulative incidence of PF as first events for BT time >56 days versus ≤56 days was 26% versus 9% (P = .04; Fig. 1). Other factors found to be associated with PF were younger age and lower baseline hemoglobin (Hgb) level. On multivariate analysis for PF (Table 4), after controlling for age and Hgb, BT time >56 days was significant. (hazard ratio [HR], 3.8; 95% CI, 1.2-16; P = .02). A separate multivariate analysis for PF was performed after excluding patients who received non-cisplatin regimens, and controlling for age and Hgb, BT time >56 days remained significant (HR, 3.2; 95% CI, 0.5-0.9; P = .05). With regard to DF, stage and PMB were associated with DF on univariate analysis, with stage remaining significant on multivariate analysis. With regard to DSM, age and stage were associated with DSM on univariate analysis only.
|Variables||3-y PF (P)||3-y DF (P)||3-y DSM (P)|
|Age, year: <49 vs ≥49||27% vs 11%||28% vs 21%||32% vs 22%|
|RT: IMRT vs non-IMRT||21% vs 14%||22% vs 28%||27% vs 27%|
|FIGO stage: I and II vs III||17% vs 23%||17% vs 47%||24% vs 37%|
|Chemo: CDDP vs alternative||17% vs 17%||24% vs 19%||27% vs 30%|
|Pelvic LN status: N0 vs N1||17% vs 22%||23% vs 34%||24% vs 37%|
|BT time: ≤56 vs >56 days||9% vs 26%||28% vs 22%||26% vs 29%|
|Total RT time: ≤56 vs >56 days||5% vs 20%||28% vs 26%||29% vs 29%|
|PMB: Yes vs No||22% vs 12%||31% vs 16%||32% vs 19%|
|Baseline Hgb: ≤10 vs >10 ng/mL||31% vs 13%||25% vs 21%||36% vs 24%|
|Tumor size: ≤5.5 vs >5.5 cm||18% vs 24%||23% vs 28%||26% vs 34%|
|Variables||Hazard Ratio||95% Confidence Interval||P|
|PF||Age (<49 vs ≥49)||2.9||1.0 to 9.3||0.05|
|Pretreatment Hgb (>10 vs ≤10 ng/mL)||0.49||0.18 to 1.4||0.2|
|Brachytherapy time (>56 vs ≤56 d)||3.8||1.2 to 16||0.02|
|DF||Parametrial boost (yes vs. no)||1.8||0.8 to 5.2||0.2|
|Stage III vs I and II||3.4||1.5 to 7.7||0.005|
|DSM||Age (<49 vs ≥49)||1.9||0.9 to 4.2||0.1|
|Stage III vs I and II||2.0||0.9 to 4.3||0.1|
We further compared clinical, treatment, and toxicity characteristics of patients who completed BT >56 days versus ≤56 days to explore reasons for treatment prolongation (Table 5). In terms of baseline characteristics, age, race, treatment year, nodal status, histology, and baseline Hgb level were well-balanced between the 2 groups. Percentage of patients who received PMB, IMRT, and CDDP-based chemotherapy were also well-balanced between the 2 groups. The only imbalance in adverse features that we found was that patients with BT >56 days had slightly larger tumors. In terms of treatment characteristics, patients with BT ≤56 days had shorter RT intervals, including WP RT duration and WP RT-BT duration. In addition, they were more likely to have LDR BT and have BT to be performed in 1 fraction. In terms of toxicities, occurrences of grade 2 and greater (grade 2+) and grade 3+ toxicities were reported in 50% and 7% of all patients, respectively. Grade 3+ toxicities were hematologic in 3 patients, gastrointestinal in 3 patients, and dermatologic in 2 patients. Although grade 2+ toxicities were well-balanced between the 2 groups, patients with BT >56 days had a significantly higher incidence of grade 3 toxicities.
|Characteristics||≤56 days (n = 49)||>56 days (n = 64)||P|
|Age median, y||49||49||NS|
|Year of treatment||NS|
|≤2000||13 (27)||19 (30)|
|2001-2004||13 (27)||16 (25)|
|2005-2008||18 (37)||24 (38)|
|≥2009||5 (10)||5 (8)|
|Nonwhite race, no. (%)||33 (67)||40 (64)||NS|
|IB2||10 (20)||11 (17)|
|II||28 (57)||37 (58)|
|III||11 (22)||15 (25)|
|Pelvic LN status||NS|
|N0||36 (73)||46 (72)|
|N1||6 (12)||13 (20)|
|Nx||7 (14)||5 (8)|
|Nonsquamous histology||4 (8)||4 (6)||NS|
|IMRT, no. (%)||27 (55)||39 (61)||NS|
|Median tumor size (cm)||5||6||0.009|
|≤5.5 cm, no.(%)||26 (53)||23 (36)||0.08|
|Median baseline Hgb (ng/mL)||11.7||11.5||NS|
|≤10 ng/mL||11 (22)||15 (23)||NS|
|WP RT time, median (d)||35||36||0.005|
|WP RT-BT interval, median (d)||13||24||<0.0001|
|Total RT time, median (d)||58||75||<0.0001|
|LDR BT, no. (%)||49 (100)||59 (92)||0.02|
|LDR BT performed in 1 fraction, no. (%)a||47 (96)||44 (75)||0.005|
|Any grade 2+ acute toxicities, no. (%)||29 (59)||28 (44)||0.1|
|Any grade 3+ acute toxicities, no. (%)||1 (2)||7 (11)||0.05|
|Parametrial boost, no. (%)||29 (59)||36 (56)||NS|
|CDDP-based chemotherapy, no. (%)||44 (90)||57 (89)||NS|
Several retrospective studies have reported poorer pelvic control and cancer-specific survival in patients whose treatment times are prolonged.6-9 However, these reports predominantly predate the era of CCRT as the standard treatment for locally advanced cervical carcinoma. One study by Chen et al included a small proportion of patients treated with concurrent chemoradiotherapy.9 The effect of concurrent chemotherapy on treatment delay has not been well studied. Nugent et al from Washington University, St. Louis, Missouri, reported a correlation between poorer PFS and OS with longer time to RT completion among a cohort of cervical cancer patients treated with CCRT.18 However, treatment delay was undefined in the study and outcome data was not reported. Another retrospective analysis of women treated with weekly CDDP and pelvic RT on the GOG 165 protocol found that treatment delay, defined as total RT ≥ 8 weeks, was associated with worse PFS and OS.19 However, RT delay was not entered into the multivariate analysis, and the patterns of failure were not reported.
Our study aimed to explore whether treatment time in patients who received CCRT was associated with PF, DF, or DSM. By examining separately times to completion of WP RT, BT, and PMB, we were also able to make associations with effect on pelvic control and distant control. A period of 56 days was used to stratify treatment time, per the recommendation of current RTOG/GOG protocols.10, 11 We found that BT time was significantly associated with PF. Three-year PF for BT time >56 days versus ≤56 days was 26% versus 9% (P = .04). This is consistent with a model in which prolonged RT time allows for accelerated repopulation of malignant cells that cannot be completely eradicated with concurrent chemotherapy, thus leading to lower local tumor control. In addition, our study confirmed that younger age or lower pretreatment Hgb were associated with higher PF on univariate analysis, with age remaining a significant factor on multivariate analysis. These findings are consistent with other studies.20, 21
Interestingly, longer BT times and total RT times were not associated with higher DF or DSM. This may indicate that disease present in the radiated field is not necessarily the nidus for distant disease and may speak to the positive effects of chemotherapy on distant disease. However, the lack of effect may also be due to the small numbers of patients in our study to detect a true difference. With longer follow-up and a larger cohort size, the effects of longer treatment time analyzed separately by BT time and total RT time might elucidate effects on DSM and indicate whether patients are adversely affected by an increase in pelvic failure.
In our study, the median total RT time was 68 days, which is longer than other retrospective series, which range from 51 to 65 days.7, 8, 19 We can make a few conclusions regarding the reasons for treatment prolongation based on our findings (Table 5), which compared the characteristics of patients with and without treatment prolongation. Reasons for treatment prolongation were multifactorial; however, treatment prolongation is likely directly related to PF and not simply a surrogate for adverse factors. A larger tumor in itself could not have explained the higher PF in patients with BT >56 days. On univariate analysis (Table 3), patients with tumors >5.5 cm had slightly worse PF than patients with tumors ≤5.5 cm, but the difference did not reach statistical significance. A separate multivariate analysis was performed that included tumor size in the model for PF, and BT >56 days remained significant after adjusting for age and baseline hemoglobin level (HR, 4.2; 95% CI, 1.3-18.9; P = .02). Larger tumors, however, might necessitate more than 1 BT fraction when using LDR in order to achieve acceptable dosimetry. Because total treatment time was increased when BT was delivered in more than 1 fraction, there might also be an indirect relationship between larger tumors and treatment prolongation. HDR BT was usually delivered in 5 or 6 fractions, and therefore, it was also found to be associated with BT >56 days in our study. Although incidence of grade 3+ acute toxicities was associated with BT >56 days, on univariate analysis, incidence of grade 3+ acute toxicities was not significantly associated with PF (data not shown), most likely because there were other modulating factors that affect outcome, such as characteristics of individual patient/tumor and a physician's threshold for giving RT breaks for toxicities. Because of the low rates (7%) of overall grade 3+ toxicity in our series, it is unlikely that the established benefit of concurrent chemotherapy was lost due to side effects.
In our analysis, PF was associated with a prolonged time to complete BT (BT time). We found that a prolonged time to complete BT was primarily attributable to a prolonged interval between the completion of pelvic radiation and institution of BT and to grade 3+ toxicities during treatment. These findings highlight the importance for instituting retrospective departmental QA to periodically examine treatment time and delineate areas for improvement. We believe that treatment prolongation might be prevented with timely BT planning. In our experience, BT planning requires joint efforts from the treating physician, the RN, and the social workers. Operating room and physician availability, patient transportation, and effective patient communication and education are all factors that could affect BT time and thus need to be coordinated. Although not captured by this analysis, patients' performance status and delay in preoperative clearance by anesthesia might also affect BT time. In addition, prompt recognition and treatment of acute toxicities might prevent significant treatment prolongation. Furthermore, although the grading of toxicities was objective, the decisions to give RT breaks were subjective. This analysis might raise clinicians' awareness of the adverse effects of interrupting radiation treatment and thus change their thresholds at which to offer an RT break.
Despite the longer treatment times in our study, the survival outcome and PF rate of our patients are comparable to those patients treated on the CCRT arm from the randomized, controlled trials with 3-year OS ranging from 67% to 75%, 3-year DFS of 59% to 69%, and 3-year PF of 15% to 25%.2, 4, 5 The survival of the patients with stage IB2 disease was unexpectedly lower than that of patients with stage II disease in our study. Comparison of baseline and treatment characteristics of stage IB2 and patients with stage II disease revealed that they had similar age, baseline Hgb, BT time, IMRT use, and nodal status. However, patients with stage IB2 vs stage II disease had slightly larger tumors (6 cm vs 5.5 cm, respectively; P = .05), and a higher percentage of patients who were not staged with CT scan (24% vs 11%, respectively; P < .0001). The combination of larger tumor size and higher uncertainty of the nodal status of patients with stage IB2 disease could explain their inferior outcome. Because of the very small numbers, it is also plausible the inferior outcome is due to chance.
We acknowledge the limitations of this study, including the retrospective nature of the review, small number of patients, and relatively short follow-up period. Reasons for the short follow-up were multifactorial. Patients may have had limited follow-up at our institutions because they are tertiary centers. Because 90% of the recurrences occurred within 1.5 years after the completion of treatment and no recurrences were reported beyond 3 years of follow-up, we feel that the relatively short follow-up times do not compromise the analysis for PF or DF, which is the primary aim of this article. Despite these limitations, our results indicate that longer time to completion of BT is associated with higher PF with no effect on distant control or DSM. Efforts to minimize treatment prolongation should focus on delivering BT in a timely fashion and management of acute toxicities.
No specific funding was disclosed.
CONFLICT OF INTEREST DISCLOSURE
The authors made no disclosure.