Abstract
This study examines factors associated with prolonged radiation duration and its impact on survival in women with cervical cancer treated with primary chemoradiation. Women in the National Cancer Database with stage IB2-IVA cervical cancer from 2003-2011 who received radiation and chemotherapy were included. Of 7209 women who met inclusion criteria, 3401 (47.1%) and 3808 (52.8%) completed radiation in ≤8 and >8 weeks, respectively. There was no overall survival difference for radiation duration ≤8 vs. >8 weeks. Sensitivity analyses showed that inferior overall survival is only seen with radiation duration of >10 to 12 weeks.
Introduction
Contemporary treatment for locally advanced cervical cancer, commonly defined as stage IB2 to IVA disease, consists of radiotherapy and chemotherapy (chemoradiation). Radiotherapy includes external beam radiation to the whole pelvis, along with brachytherapy delivered locally to the cervix. Concurrent chemoradiation became the standard of care in 1999 with the publication of 5 clinical trials demonstrating that the addition of chemotherapy to the radiation regimen confers a substantial survival benefit [1-5].
Several retrospective, mostly small, single-institution studies have demonstrated that the duration of radiotherapy can negatively affect cervical cancer outcomes; prolonging the radiation course in these studies was associated with a decrease of 0.5% to 1.2% in pelvic control and cause-specific survival per additional day of treatment [6-11]. The adverse effects of prolonged radiation have been demonstrated in head and neck cancers as well [11-15]. A possible physiologic mechanism for these adverse outcomes is accelerated tumor cell repopulation during radiotherapy when there is a lag period after initiation of radiation [12].
The National Comprehensive Cancer Network (NCCN) Cervical Cancer Guidelines recommend that chemoradiation be completed within 8 weeks [16], and this is being proposed as a quality measure for cervical cancer treatment. However, this recommendation is based on the aforementioned studies that were published prior to the incorporation of chemotherapy in the treatment of cervical cancer. The literature on the effect of radiation duration in the era of concurrent chemotherapy is limited. Recently, two single-institution retrospective studies demonstrated that prolonged chemoradiation duration did not negatively impact survival [17,18]. We performed a population-based analysis to determine factors associated with prolonged chemoradiation duration and to assess its impact on overall (not cause-specific) survival in women with locally advanced cervical cancer treated with primary chemoradiation.
Materials and Methods
Study Design and Data Source
We utilized data from the National Cancer Database (NCDB). The NCDB, established in 1989, is a joint project of the American Cancer Society and the Commission on Cancer (CoC) of the American College of Surgeons (ACS). NCDB collects data from over 1,500 hospitals in the US with Commission-accredited cancer programs, capturing 70% of all newly diagnosed cancers in the US per year. Data pertaining to the initial diagnosis and first course of treatment are collected and submitted to the NCDB using nationally standardized data item and coding definitions [19]. Participating registries report follow-up of vital status to the NCDB on a yearly basis. An annual 90% follow-up rate for all patients diagnosed within the preceding 5 years and an 80% follow-up rate for all other cases is required.
Cohort Selection
All patients diagnosed with stage IB2 to IVA cervical cancer diagnosed from 2003 to 2011 who received combination external beam radiation therapy and brachytherapy with concurrent chemotherapy were included (Figure 1). Women were excluded if they had more than one reported cancer diagnosis. We did not include women who were diagnosed prior to 2003 because the variable describing radiation treatment in the database was reported as an optional item, and comorbidity data were not collected until the year 2003. Women were excluded if they underwent a definitive surgical procedure, such as a trachelectomy, simple or radical hysterectomy, or exenteration. Additionally, since our primary outcome of interest and predictor of survival was prolonged radiation, we focused our analysis on women who completed at least 6 weeks of radiation. We excluded women with radiation treatment duration greater than 26 weeks, since such cases represent extremely rare clinical scenarios and/or spurious data collection.
Figure 1.
CONSORT diagram detailing cohort selection for analysis.
Variable Definition
Radiation duration is defined as the number of days between the radiation start and end date, categorized into number of weeks. Age was modeled as a continuous variable. Race was categorized as non-Hispanic white, non-Hispanic black, Hispanic, Asian/other, and unknown. Insurance status was categorized as private, Medicare, Medicaid, uninsured, and other/unknown. The Charlson/Deyo comorbidity score (CDS), commonly used in health outcomes studies employing administrative databases, provides an indicator of various comorbid conditions based on reported ICD-9-CM codes [19, 20]. In the NCDB, the scores have been classified as 0, 1, or 2. Year of diagnosis was modeled as a categorical variable with each year as a separate category. Clinical and pathologic data, including tumor histology, grade, and AJCC clinical stage were recorded. Vital status and the number of months between the date of diagnosis and date of last contact were also recorded. Patients who were alive at last contact were censored.
Environmental and structural factors, including reporting facility type, reporting facility geographic location, woman's neighborhood level median income, neighborhood level educational attainment, and neighborhood-level measure of rurality/urbanicity were recorded. Reporting facility type is assigned in the NCDB as an Academic Program, Community Cancer Program, Comprehensive Community Cancer Program, and other specified types of cancer programs. Reporting facility location was consolidated into four categories based on US Census Divisions: Northeast, Midwest, South, and West. Neighborhood-level median income and educational attainment are derived in the NCDB by linking the woman's zip code at the time of diagnosis to US Census data files from the year 2000; therefore these data represent the zip-code of the woman's residence and not that of the individual woman. Median household income and educational attainment were categorized as quartiles based on equally proportioned income and educational attainment ranges for all US zip codes. The neighborhood-level measure of rurality/urbanicity was obtained using the typology published by the USDA Economic Research Service.
Statistical analysis
Radiation duration was modeled as a dichotomous variable, with prolonged duration defined as greater than 8 weeks for our primary analysis. Frequency distributions between categorical variables by radiation duration were compared using chi-squared tests. Binomial logistic regression analysis was used to determine factors associated with receipt of prolonged radiation, including age, race, stage, histology, grade, year of diagnosis, CDS, income, education, rurality/urbanicity, facility type and facility location. Outcomes are reported as odds ratios with 95% confidence intervals. In accordance with the NCDB βParticipant Use File Guide, survival analysis was limited to cases diagnosed from 2003 to 2006 to allow for at least 5 years of follow-up time. Kaplan Meier survival analysis with the log-rank test for statistical significance was performed to assess impact of prolonged therapy on survival. Cause of death is not reported in the NCDB, so cause-specific survival could not be calculated. Cox proportional hazards models were used to examine the impact of prolonged radiation on risk of death, adjusted by age, race, stage, histology, grade, year of diagnosis, CDS, education, rurality/urbanicity, facility type and facility location.
Covariates included in the regression models were selected a priori based on clinical relevance. We found evidence of multi-collinearity between education and income, and therefore did not include both measures in the final models. The Akaike Information Criterion value [21] for the logistic regression model including education was slightly better than the model including income, and since education may be a stronger indicator of socioeconomic status [22], income was not included in the final models. In order to explore effect modification by stage, we performed stage-stratified analyses for early (stages I/II) and late stage (stages III/IV) tumors were performed. We also performed sensitivity analyses using additional cut-offs for prolonged radiation duration: ≤ or > 9, 10, and 12 weeks. Statistical significance was set to a p-value of less than 0.05. All statistical tests were two-sided and all analyses were performed using SAS version 9.4 (SAS Institute, Inc, Cary, NC).
Results
Of the 51,181 women with stage IB2 – IVA cervical cancer reported in the NCDB from 1998 to 2011, 9905 met initial inclusion criteria of having been treated with external beam radiation, brachytherapy, and chemotherapy. After applying the exclusion criteria, the final cohort included 7209 women (Figure 1). Of these, 3401 (47.1%) and 3808 (52.8%) completed radiation in 8 weeks or less and greater than 8 weeks, respectively. Figure 2 shows the distribution of radiation duration by number of weeks, which demonstrates a sharp decline in the number of women with radiation duration greater than 9 to 10 weeks. Age, race, CDS, clinical stage, tumor histology and size, insurance status, education, facility location, and neighborhood rurality/urbanicity were all associated with prolonged radiation duration on bivariate analysis, as shown in Table 1 (p<0.05 for all). A time trend was seen by diagnosis year, where women diagnosed in years 2003 to 2007 were more likely to have prolonged radiation duration, and those diagnosed after 2007 were more likely to have radiation duration less than 8 weeks (p=0.001).
Figure 2.
Distribution of Radiation Duration in the Study Cohort by Number of Weeks, NCDB 2003 – 2011.
Table 1.
Association between radiotherapy duration and characteristics of women with stage IB2-IVA cervical cancer treated with primary chemoradiation, NCDB, 2003 – 2011.
| Characteristic | All women N = 7209 | RT ≤ 8 weeks N = 3401 | RT > 8 weeks N = 3808 | P* | |||
|---|---|---|---|---|---|---|---|
| No. | % | No. | % | No. | % | ||
| Age in years (mean, standard deviation) | 51.0 (12.9) | - | 50.8 (12.9) | - | 51.3 (13.0) | - | 0.101 |
| Race | 0.002 | ||||||
| White, non-Hispanic | 4551 | 63.1 | 2204 | 64.8 | 2347 | 61.6 | |
| Black, non-Hispanic | 1242 | 17.2 | 528 | 15.5 | 714 | 18.8 | |
| Hispanic | 941 | 13.1 | 434 | 12.8 | 507 | 13.3 | |
| Asian/Other | 403 | 5.6 | 205 | 6.0 | 198 | 5.2 | |
| Unknown | 72 | 1.0 | 30 | 0.9 | 42 | 1.1 | |
| Charlson/Deyo Comorbidity Index | 0.006 | ||||||
| 0 | 6362 | 88.3 | 3044 | 89.5 | 3318 | 87.1 | |
| 1 | 708 | 9.8 | 302 | 8.9 | 406 | 10.7 | |
| 2 | 139 | 1.9 | 55 | 1.6 | 84 | 2.2 | |
| Year of Diagnosis | 0.001 | ||||||
| 2003 | 596 | 8.3 | 279 | 8.2 | 317 | 8.3 | |
| 2004 | 699 | 9.7 | 314 | 9.2 | 385 | 10.1 | |
| 2005 | 724 | 10.0 | 295 | 8.7 | 429 | 11.3 | |
| 2006 | 765 | 10.6 | 346 | 10.2 | 419 | 11.0 | |
| 2007 | 835 | 11.6 | 387 | 11.4 | 448 | 11.8 | |
| 2008 | 906 | 12.6 | 435 | 12.8 | 471 | 12.4 | |
| 2009 | 896 | 12.4 | 439 | 12.9 | 457 | 12.0 | |
| 2010 | 836 | 11.6 | 412 | 12.1 | 424 | 11.1 | |
| 2011 | 952 | 13.2 | 494 | 14.5 | 458 | 12.0 | |
| Clinical Stage | <0.001 | ||||||
| IB2 | 609 | 8.5 | 362 | 10.6 | 247 | 6.5 | |
| IINOS | 109 | 1.5 | 61 | 1.8 | 48 | 1.3 | |
| IIA | 484 | 6.7 | 263 | 7.7 | 221 | 5.8 | |
| IIA1 | 23 | 0.3 | 11 | 0.3 | 12 | 0.3 | |
| IIA2 | 47 | 0.6 | 23 | 0.7 | 24 | 0.6 | |
| IIB | 2547 | 35.3 | 1240 | 36.5 | 1307 | 34.3 | |
| IIINOS | 86 | 1.2 | 36 | 1.1 | 50 | 1.3 | |
| IIIA | 173 | 2.4 | 76 | 2.2 | 97 | 2.6 | |
| IIIB | 2840 | 39.4 | 1219 | 35.8 | 1621 | 42.6 | |
| IVNOS | 67 | 0.9 | 32 | 0.9 | 35 | 0.9 | |
| IVA | 224 | 3.1 | 78 | 2.3 | 146 | 3.8 | |
| Tumor Size | <0.001 | ||||||
| <20 mm | 116 | 1.6 | 67 | 2.0 | 49 | 1.3 | |
| 20-40 mm | 824 | 11.4 | 402 | 11.8 | 422 | 11.1 | |
| 41-60 mm | 1941 | 26.9 | 992 | 29.2 | 949 | 24.9 | |
| 61-80 mm | 1222 | 16.9 | 563 | 16.6 | 659 | 17.3 | |
| >80 mm | 340 | 4.7 | 156 | 4.6 | 184 | 4.8 | |
| Unknown | 2766 | 38.4 | 1221 | 35.9 | 1545 | 40.6 | |
| Tumor Histology | 0.010 | ||||||
| Squamous | 5876 | 81.5 | 2738 | 80.5 | 3138 | 82.4 | |
| Adenocarcinoma | 628 | 8.7 | 320 | 9.4 | 308 | 8.1 | |
| Adenosquamous | 223 | 3.1 | 124 | 3.7 | 99 | 2.6 | |
| Other | 482 | 6.7 | 219 | 6.4 | 263 | 6.9 | |
| Tumor Grade | 0.953 | ||||||
| I | 304 | 4.2 | 145 | 4.3 | 159 | 4.2 | |
| II | 2293 | 31.8 | 1077 | 31.7 | 1216 | 31.9 | |
| III | 2416 | 33.5 | 1150 | 33.8 | 1266 | 33.3 | |
| Unknown | 2196 | 30.5 | 1029 | 30.3 | 1167 | 30.6 | |
| Insurance | 0.003 | ||||||
| Private | 3055 | 42.4 | 1497 | 44.0 | 1558 | 40.9 | |
| Medicare | 1177 | 16.3 | 539 | 15.9 | 638 | 16.7 | |
| Medicaid | 1915 | 26.6 | 855 | 25.1 | 1060 | 27.8 | |
| Not insured | 856 | 11.9 | 395 | 11.6 | 461 | 12.1 | |
| Other/Unknown | 206 | 2.9 | 115 | 3.4 | 91 | 2.4 | |
| Median Neighborhood Education | 0.012 | ||||||
| ≥ 29% without high school diploma | 1791 | 24.8 | 789 | 23.2 | 1002 | 26.3 | |
| 20 – 28.9% | 1939 | 27.0 | 923 | 27.1 | 1016 | 26.7 | |
| 14 – 19.9% | 1530 | 21.2 | 740 | 21.8 | 790 | 20.7 | |
| < 14% | 1601 | 22.2 | 765 | 22.5 | 836 | 22.0 | |
| Unknown | 348 | 4.8 | 184 | 5.4 | 164 | 4.3 | |
| Median Neighborhood Income | 0.095 | ||||||
| < $30,000 | 1366 | 19.0 | 613 | 18.0 | 753 | 19.8 | |
| $30,000 - $34,999 | 1510 | 20.9 | 717 | 21.1 | 793 | 20.8 | |
| $35,000 - $45,999 | 2031 | 28.2 | 952 | 28.0 | 1079 | 28.3 | |
| ≥ $46,000 | 1954 | 27.1 | 935 | 27.5 | 1019 | 26.7 | |
| Unknown | 348 | 4.8 | 184 | 5.4 | 164 | 4.3 | |
| Facility Type | 0.712 | ||||||
| Academic | 3555 | 49.3 | 1692 | 49.7 | 1863 | 48.9 | |
| Comprehensive Community Cancer | 3172 | 44.0 | 1474 | 43.3 | 1698 | 44.6 | |
| Community Cancer | 459 | 6.4 | 224 | 6.6 | 235 | 6.2 | |
| Other | 23 | 0.3 | 11 | 0.3 | 12 | 0.3 | |
| Facility Location | <0.001 | ||||||
| Northeast | 1497 | 20.8 | 638 | 18.7 | 859 | 22.6 | |
| South | 2526 | 35.0 | 1156 | 34.0 | 1370 | 36.0 | |
| West | 1262 | 17.5 | 646 | 19.0 | 616 | 16.2 | |
| Midwest | 1924 | 26.7 | 961 | 28.3 | 963 | 25.3 | |
| Neighborhood Rurality/Urbanicity | 0.019 | ||||||
| Metro | 5538 | 76.8 | 2608 | 76.7 | 2930 | 76.9 | |
| Urban | 1164 | 16.2 | 567 | 16.7 | 597 | 15.7 | |
| Rural | 119 | 1.6 | 40 | 1.2 | 79 | 2.1 | |
| Unknown | 388 | 5.4 | 186 | 5.5 | 202 | 5.3 | |
Abbreviations: RT = radiation
P value given for ANOVA for continuous variables and chi-square tests for categorical variables.
On multivariable analysis (Table 2), black race remained a significant factor associated with prolonged radiation, with black women being 20% more likely to receive prolonged radiation compared to white women (OR 1.20, 95% CI 1.04 – 1.37). Other racial/ethnic groups had similar odds of receiving prolonged radiation as white women. Women with CDS of 1 and 2 were also more likely to have prolonged radiation duration relative to those with CDS of 0, although this was only statistically significant for women with a score of 1 (CDS 1: OR 1.19, 95% CI 1.01 – 1.40; CDS 2: OR 1.33, 95% CI 0.96 – 1.89). Clinical stage greater than IIA2 was found to be significantly associated with prolonged radiation. Medicaid recipients were 19% more likely to receive prolonged radiation compared to women with private insurance (OR 1.19, 95% CI 1.06 – 1.34), whereas women with other/unknown insurance status were less likely (OR 0.73, 95% CI 0.55 – 0.98) to have a prolonged course of radiotherapy. Women living in the Northeast and the South were 36% (OR 1.36, 95% CI 1.18 – 1.56) and 17% (OR 1.17, 95% CI 1.04 – 1.33) more likely to get prolonged radiation therapy, respectively, than those living in the Midwest. Residents of rural neighborhoods were 78% more likely to have prolonged radiation therapy than those in urban neighborhoods (OR 1.78, 95% CI 1.20 – 2.64).
Table 2.
Multivariable analysis of factors associated with receipt of prolonged radiation (≤ and >8 weeks) for women with stage IB2-IVA cervical cancer treated with primary chemoradiation, NCDB, 2003 – 2011.
| Adjusted* Odds Ratio | 95% CI | P value | |
|---|---|---|---|
| Age | 1.00 | 1.00 – 1.01 | 0.566 |
| Race | |||
| White, non-Hispanic | Referent | - | - |
| Black, non-Hispanic | 1.20 | 1.04 – 1.37 | 0.010 |
| Hispanic | 1.11 | 0.95 – 1.30 | 0.182 |
| Asian/Other | 0.97 | 0.78 – 1.20 | 0.774 |
| Unknown | 1.41 | 0.87 – 2.28 | 0.159 |
| Charlson/Deyo Comorbidity Index | |||
| 0 | Referent | - | - |
| 1 | 1.19 | 1.01 – 1.40 | 0.034 |
| 2 | 1.33 | 0.94 – 1.89 | 0.111 |
| Year of Diagnosis | |||
| 2003 | Referent | - | - |
| 2004 | 1.10 | 0.88 – 1.38 | 0.393 |
| 2005 | 1.30 | 1.04 – 1.62 | 0.021 |
| 2006 | 1.09 | 0.87 – 1.35 | 0.450 |
| 2007 | 1.02 | 0.82 – 1.26 | 0.889 |
| 2008 | 0.96 | 0.78 – 1.18 | 0.710 |
| 2009 | 0.92 | 0.74 – 1.13 | 0.412 |
| 2010 | 0.90 | 0.72 – 1.11 | 0.321 |
| 2011 | 0.81 | 0.66 – 1.00 | 0.048 |
| Clinical Stage | |||
| IB2 | Referent | - | - |
| IINOS | 1.15 | 0.76 – 1.75 | 0.498 |
| IIA | 1.19 | 0.91 – 1.49 | 0.220 |
| IIA1 | 1.82 | 0.78 – 4.24 | 0.166 |
| IIA2 | 1.84 | 1.01 – 3.38 | 0.048 |
| IIB | 1.47 | 1.22 – 1.77 | <0.001 |
| IIINOS | 1.90 | 1.20 – 3.03 | 0.007 |
| IIIA | 1.79 | 1.27 – 2.54 | 0.001 |
| IIIB | 1.88 | 1.57 – 2.26 | <0.001 |
| IVNOS | 1.58 | 0.94 – 2.63 | 0.081 |
| IVA | 2.70 | 1.95 – 3.73 | <0.001 |
| Tumor Histology | |||
| Squamous | Referent | - | - |
| Adenocarcinoma | 0.93 | 0.79 – 1.11 | 0.425 |
| Adenosquamous | 0.71 | 0.54 – 0.93 | 0.013 |
| Other | 1.07 | 0.88 – 1.29 | 0.511 |
| Tumor Grade | |||
| I | Referent | - | - |
| II | 1.02 | 0.80 – 1.31 | 0.860 |
| III | 1.00 | 0.78 – 1.28 | 0.999 |
| Unknown | 1.05 | 0.82 – 1.35 | 0.688 |
| Insurance | |||
| Private | Referent | - | - |
| Medicare | 1.06 | 0.91 – 1.25 | 0.456 |
| Medicaid | 1.19 | 1.05 – 1.34 | 0.005 |
| Not insured | 1.11 | 0.94 – 1.30 | 0.209 |
| Other/Unknown | 0.73 | 0.55 – 0.98 | 0.035 |
| Median Education Attainment | |||
| ≥ 29% without high school diploma | Referent | - | - |
| 20 – 28.9% | 0.89 | 0.78 – 1.02 | 0.095 |
| 14 – 19.9% | 0.92 | 0.79 – 1.06 | 0.248 |
| < 14% | 0.95 | 0.82 – 1.10 | 0.478 |
| Unknown | 0.63 | 0.45 – 0.87 | 0.005 |
| Facility Type | |||
| Academic | Referent | - | - |
| Comprehensive Community Cancer | 1.08 | 0.98 – 1.20 | 0.118 |
| Community Cancer | 0.96 | 0.79 – 1.17 | 0.711 |
| Other | 0.98 | 0.42 – 2.26 | 0.953 |
| Facility Location | |||
| Midwest | Referent | - | - |
| Northeast | 1.35 | 1.18 – 1.56 | <0.001 |
| South | 1.17 | 1.04 – 1.33 | 0.012 |
| West | 0.94 | 0.81 – 1.10 | 0.446 |
| Neighborhood Rurality/Urbanicity | |||
| Metro | Referent | - | - |
| Urban | 0.98 | 0.86 – 1.12 | 0.761 |
| Rural | 1.78 | 1.20 – 2.64 | 0.004 |
| Unknown | 1.26 | 0.93 – 1.70 | 0.130 |
Abbreviations: CI = confidence interval; UL = upper limit; LL = lower limit
Adjusted for age, race, stage, histology, grade, year of diagnosis, Charlson/Deyo score, education, rurality/urbanicity, and facility type and location.
There was no difference in survival seen between women with radiation duration of greater than 8 weeks vs. 8 or fewer weeks (p=0.209) (Figure 2). In the series of sensitivity analyses in which varying cut points for duration of radiation were examined, inferior survival was seen with radiotherapy duration greater than 9 weeks (p=0.001), 10 weeks (p<0.001), and 12 weeks (p=<0.001).
Our primary Cox proportional hazards model included radiation duration dichotomized using an 8-week cut-point (Table 3) and showed no association between radiation duration of >8 weeks and survival. Effect modification by stage was explored by stratifying the Cox model. There was no difference in risk of death with prolonged radiation duration in either stage group (stages I/II: HR 0.88, 95% CI 0.74 – 1.04; stages III/IV: HR 1.10, 95% CI 0.93 – 1.28).
Table 3.
Adjusted Cox Proportional Hazards Model analysis of survival for women with stage IB2-IVA cervical cancer treated with primary chemoradiation, NCDB, 2003 – 2006.
| No per Group | Adjusted* Hazards Ratio | 95% CI | P value | |
|---|---|---|---|---|
| Radiation duration, all stages | ||||
| ≤ 8 weeks | 1234 | Referent | Referent | Referent |
| > 8 weeks | 1550 | 0.98 | 0.87 – 1.10 | 0.719 |
| Radiation duration, stages I and II | ||||
| ≤ 8 weeks | 751 | Referent | Referent | Referent |
| > 8 weeks | 794 | 0.88 | 0.74 – 1.04 | 0.132 |
| Radiation duration, stages III and IV | ||||
| ≤ 8 weeks | 483 | Referent | Referent | Referent |
| > 8 weeks | 756 | 1.10 | 0.93 – 1.28 | 0.265 |
| Radiation duration, all stages | ||||
| Continuous, # of days | - | 1.01 | 0.99 – 1.03 | 1.011 |
| Radiation duration, all stages | ||||
| ≤ 9 weeks | 1751 | Referent | Referent | Referent |
| > 9 weeks | 1033 | 1.12 | 1.00 – 1.26 | 0.055 |
| Radiation duration, all stages | ||||
| ≤ 10 weeks | 2166 | Referent | Referent | Referent |
| > 10 weeks | 668 | 1.15 | 1.01 – 1.30 | 0.037 |
| Radiation duration, all stages | ||||
| ≤ 12 weeks | 2503 | Referent | Referent | Referent |
| > 12 weeks | 281 | 1.23 | 1.04 – 1.47 | 0.018 |
Abbreviations: CI = confidence interval; UL = upper limit; LL = lower limit
Results shown are from seven separate models. All models adjusted for age, race, stage, histology, grade, year of diagnosis, Charlson/Deyo score, education, rurality/urbanicity, and facility type and location.
In additional exploratory Cox models, there was no evidence that the risk of death increased per additional day of radiation duration (HR 1.01, 95% CI 0.99 – 1.03), and no difference in risk of death with a 9-week cut-point (HR 1.12, 95% CI 1.00 – 1.26). However, women with duration of greater than 10 weeks had a 15% higher risk of death compared to those who completed radiation in less than 10 weeks (HR 1.15, 95% CI 1.01 – 1.30). Women with duration of greater than 12 weeks had a 23% higher risk of death compared to those who completed radiation in less than 12 weeks (HR 1.23, 95% CI 1.04 – 1.47). Since there was a difference in survival seen with a 10-week cut-off, we explored factors associated with radiation duration of greater than 10 weeks using logistic regression. The results were similar for the logistic regression analysis using an 8-week cut-off. Black race, Medicaid insurance, living in the Northeast and the South, and living in rural neighborhoods remained significant factors associated with prolonged radiation duration. In addition, Hispanic women and those who were treated in Community and Comprehensive Community Cancer Programs were also found to be more likely to receive radiation duration (see supplemental table).
Discussion
Our findings demonstrate that prolonging the course of radiation up to 10 weeks has no effect on survival when chemotherapy is given concurrently. Sociodemographic factors are important predictors of radiotherapy duration for cervical cancer. This study provides strong evidence against the use of radiation duration of less than 8 weeks as a quality metric for cervical cancer in the era of chemoradiation.
There is an overall paucity of data on the impact of radiation duration on survival in women treated with concurrent chemoradiation for cervical cancer. Concurrent chemotherapy has been shown to increase the biologic effective dose of radiation and reduce the accelerated repopulation of tumor cells that occurs with prolonged radiation duration [23, 24, 25]. In a study on the effect of prolonged treatment with high-dose-rate intracavitary radiation in women with cervical cancer published in 2003, the authors found that the pelvic control rate and 5-year cause-specific survival were significantly worse for those whose treatment duration exceeded 63 days. However, only 45 of 257 subjects (17.5%) received chemotherapy, and the regimen was not the current standard of weekly cisplatin [26]. Two recent studies that specifically addressed this question have been published. Song, et al. performed a single-institution retrospective study on 113 women who were treated with chemoradiation from 1997 to 2009 [17]. They found that radiation duration of greater than 8 weeks was associated with increased pelvic tumor recurrence, but not distant site recurrence or cause-specific survival. A similar single-institution retrospective study on 166 women treated with chemoradiation from 1989 to 2009 found that radiation duration of greater than 60 days was not associated with increased pelvic tumor recurrence or cause-specific survival [18]. These findings are consistent with those of our study that a cut-point of 8 weeks is not associated with inferior survival when women are treated with both chemotherapy and radiation. However, the results of these prior studies should be cautiously interpreted given their small sample size.
Several factors were found to be associated with prolonged radiation duration for cervical cancer. Later stage at diagnosis increased the odds of receiving prolonged care. Perhaps this could be explained by a higher risk of complications for those with more advanced stages. Regarding demographic factors, black women and Medicaid recipients were more likely to receive prolonged radiation. This represents a socioeconomic disparity and raises the possibility of unequal access to treatment for racial minorities and the underserved. Women who live in the Northeast or in the South, or in rural areas of the country were also more likely to have radiation duration of greater than 8 weeks. In fact, rural residence had the strongest association with receipt of prolonged radiation. This finding reflects the fact that chemoradiation entails daily travel to a treatment facility, which can be logistically burdensome to patients and their families. If distance to the treatment facility plays a role, increasing the number of facilities that offer chemoradiation may mitigate the number of women who receive prolonged radiation duration.
We comprehensively explored factors associated with prolonged radiation duration and its impact on overall survival in the era of chemoradiation. Since other national cancer databases lack information on radiation duration, a population-based analysis was not possible prior to the availability of the NCDB to outcomes researchers. Another strength conferred by the use of the NCDB includes the large sample size that provides sufficient power to examine our primary outcomes. The database also provides detailed clinical and treatment information, which allowed the analysis to be adjusted for important clinical and demographic parameters.
Our study has several limitations. First, our study is susceptible to selection bias. While the NCDB collects the majority of incident cancer cases in the US, a hospital selection bias may exist since only ACS CoC-approved hospitals contribute data. We also excluded a large number of patients due to the coding of the radiation therapy variable. However, these exclusions are likely non-differential with regards to bias since the coding of radiation duration is not likely to vary across cut-points of radiation duration. Furthermore, we believe that the sensitivity of the radiation duration coding provides confidence that these cases were actually treated with combination external beam radiation and brachytherapy. The study is also limited by the lack of information on cause-specific survival and radiation or chemotherapy toxicity. We used radiation duration as a proxy for completion of the prescribed radiotherapy dose and cycles of chemotherapy. Another limitation is that the NCDB does not capture cancer recurrence information; therefore we cannot comment on the impact of radiation duration on tumor control. We also don't have information on reasons for prolonged radiation, which is important for informing policy. Lastly, it is not possible to determine individual patient and physician preferences that may have influenced treatment allocation.
Substantial evidence demonstrates inconsistent clinical practice patterns. For example, in a large study of trends in cervical cancer treatment quality in the US, the radiation treatment duration was as follows: greater than 56 days, 63 days, and 70 days for 50%, 36%, and 15% of the 3275 patients included in the study, respectively [27]. Another large population-based study demonstrated that patients who receive their cervical cancer at facilities with a high annual volume of cervical cancer patients are more likely to receive the standard of care [28]. These treatment inconsistencies, in addition to mounting healthcare costs and waning resources, has led to regulatory agencies and other stakeholders developing quality metrics for a gamut of clinical diagnoses and specialties [29, 30]. The National Quality Forum (NQF), in conjunction with the ACS, the NCCN, and the American Society of Clinical Oncology, convened a steering committee in 2002 to address quality measures in cancer care [31]. Currently, there are 19 NQF-endorsed quality measures in breast cancer, colorectal cancer, and palliative care. Hospitals with CoC-approved cancer programs that report to the NCDB are benchmarked according to their performance on these measures [19, 31]. Given the important role that these measures will play in our healthcare system, ideally performance on these measures would result in improved patient outcomes [32, 33].
The NCCN recommends that chemoradiation be completed within 8 weeks, and chemoradiation duration of less than 60 days may be considered as a national quality measure by professional organizations. However, based on these results and those of other studies, this may not be a valid quality measure for cervical cancer. A cut-point of 10 or 12 weeks appears to be more appropriate based on our findings. Although treatment should be given in a timely manner, if there is an unavoidable delay in radiation duration, patients can be counseled that it should not negatively affect survival if completed within 10 to 12 weeks. Prospective observational studies to evaluate the relationship of sociodemographic disparities to radiation duration, the association of radiation duration on recurrence in the era of chemoradiation, and reasons for prolonged radiation are warranted.
Supplementary Material
Figure 3.
Kaplan-Meier survival analysis by dichotomous radiation duration (A) ≤ 8 vs. > 8 weeks; log rank test chi-square statistic = 1.56, P = 0.209; (B) ≤ 9 vs. > 9 weeks; log rank test chi-square statistic = 10.62, P = 0.001; (C) ≤ 10 vs. > 10 weeks; log rank chi-square statistic = 12.60, P <0.001; (D) ≤ 12 vs. > 12 weeks; log rank test chi-square statistic = 16.11, P <0.001.
Acknowledgments
None
Dr. Wright (NCI R01CA169121-01A1) and Dr. Hershman (NCI R01CA134964) are recipients of grants and Dr. Tergas is the recipient of a fellowship grant (NCI R25 CA094061-11) from the National Cancer Institute.
Footnotes
Declaration of Interests: The authors report no conflicts of interest.
References
- 1.Keys HM, Bundy BN, Stehman FB, et al. Cisplatin, radiation, and adjuvant hysterectomy compared with radiation and adjuvant hys- terectomy for bulky stage IB cervical carcinoma. N Engl J Med. 1999;340:1154–1161. doi: 10.1056/NEJM199904153401503. [DOI] [PubMed] [Google Scholar]
- 2.Morris M, Eifel PJ, Lu J, et al. Pelvic radiation with concurrent chemotherapy compared with pelvic and para-aortic radiation for high-risk cervical cancer. N Engl J Med. 1999;340:1137–1143. doi: 10.1056/NEJM199904153401501. [DOI] [PubMed] [Google Scholar]
- 3.Rose PG, Bundy BN, Watkins EB, et al. Concurrent cisplatin-based radiotherapy and chemotherapy for locally advanced cervical cancer. N Engl J Med. 1999;340:1144–1153. doi: 10.1056/NEJM199904153401502. [DOI] [PubMed] [Google Scholar]
- 4.Whitney CW, Sause W, Bundy BN, et al. Randomized comparison of fluorouracil plus cisplatin versus hydroxyurea as an adjunct to radiation therapy in stage IIB-IVA carcinoma of the cervix with negative para-aortic lymph nodes: a Gynecologic Oncology Group and Southwest Oncology Group study. J Clin Oncol. 1999;17:1339–1348. doi: 10.1200/JCO.1999.17.5.1339. [DOI] [PubMed] [Google Scholar]
- 5.Peters WA, 3rd, Liu PY, Barrett RJ, 2nd, et al. Concurrent chemo- therapy and pelvic radiation therapy compared with pelvic radiation therapy alone as adjuvant therapy after radical surgery in high-risk early-stage cancer of the cervix. J Clin Oncol. 2000;18:1606–1613. doi: 10.1200/JCO.2000.18.8.1606. [DOI] [PubMed] [Google Scholar]
- 6.Fyles A, Keane TJ, Barton M, Simm J. The effect of treatment duration in the local control of cervix cancer. Radiother Oncol. 1992;25:273–279. doi: 10.1016/0167-8140(92)90247-r. [DOI] [PubMed] [Google Scholar]
- 7.Girinsky T, Rey A, Roche B, et al. Overall treatment time in advanced cervical carcinomas: a critical parameter in treatment outcome. Int J Radiat Oncol Biol Phys. 1993;27:1051–1056. doi: 10.1016/0360-3016(93)90522-w. [DOI] [PubMed] [Google Scholar]
- 8.Lanciano RM, Pajak TF, Martz K, Hanks GE. The influence of treatment time on outcome for squamous cell cancer of the uterine cervix treated with radiation: a patterns-of-care study. Int J Radiat Oncol Biol Phys. 1993;25:391–397. doi: 10.1016/0360-3016(93)90058-4. [DOI] [PubMed] [Google Scholar]
- 9.Perez CA, Grigsby PW, Castro-Vita H, Lockett MA. Carcinoma of the uterine cervix. I. Impact of prolongation of overall treatment time and timing of brachytherapy on outcome of radiation therapy. Int J Radiat Oncol Biol Phys. 1995;32:1275–1288. doi: 10.1016/0360-3016(95)00220-S. [DOI] [PubMed] [Google Scholar]
- 10.Petereit DG, Sarkaria JN, Chappell R, et al. The adverse effect of treatment prolongation in cervical carcinoma. Int J Radiat Oncol Biol Phys. 1995;32:1301–1307. doi: 10.1016/0360-3016(94)00635-X. [DOI] [PubMed] [Google Scholar]
- 11.Keane TJ, Fyles B, O'Sullivan B, Barton M, Maki E, Simm J. The effect of treatment duration on local control of squamous carcinoma of the tonsil and carcinoma of the cervix. Seminars in Radiat Oncol. 1992;2(1):26–28. [Google Scholar]
- 12.Withers HR, Taylor JM, Maciejewski B. The hazard of accelerated tumor clonogen repopulation during radiotherapy. Acta Oncologica. 1988;27:131–146. doi: 10.3109/02841868809090333. [DOI] [PubMed] [Google Scholar]
- 13.Barton MB, Keane TJ, Gadalla T, et al. The effect of treatment time and treatment interruption on tumor control following radical radiotherapy of laryngeal cancer. Radiother Oncol. 1992;23:137–143. doi: 10.1016/0167-8140(92)90323-m. [DOI] [PubMed] [Google Scholar]
- 14.Fowler JF, Lindstrom MJ. Loss of local control with prolongation in radiotherapy. Int J Radiat Oncol Biol Phys. 1992;23:457–467. doi: 10.1016/0360-3016(92)90768-d. [DOI] [PubMed] [Google Scholar]
- 15.Fu KK, Pajak TF, Trotti A, et al. A Radiation Therapy Oncology Group (RTOG) phase III randomized study to compare hyper-fractionation and two variants of accelerated fractionation to standard fractionation radiotherapy for head and neck squamous cell carcinomas: First report of RTOG 9003. Int J Radiat Oncol Biol Phys. 2000;48:7–16. doi: 10.1016/s0360-3016(00)00663-5. [DOI] [PubMed] [Google Scholar]
- 16.The National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines v2.2013, cervical cancer. Available from: http://www.nccn.org/professionals/physician_gls/f_guidelines.asp. Accessed ….
- 17.Song S, Rudra S, Hasselle MD, Dorn PL, Mell LK, Mundt AJ, Yamada SK, Lee NK, Hasan Y. The effect of treatment time in locally advanced cervical cancer in the era of concurrent chemoradiotherapy. Cancer. 2013;119:325–31. doi: 10.1002/cncr.27652. [DOI] [PubMed] [Google Scholar]
- 18.Shaverdian N, Gondi V, Sklenar KL, Dunn EF, Petereit DG, Straub MR, Bradley KA. Effects of treatment duration during concomitant chemoradiation therapy for cervical cancer. Int J Radiation Bio Phys. 2013;86(3):562–568. doi: 10.1016/j.ijrobp.2013.01.037. [DOI] [PubMed] [Google Scholar]
- 19.Bilimoria KY, Stewart AK, Winchester DP, Ko CY. The National Cancer Data Base: A powerful initiative to improve cancer care in the United States. Ann Surg Oncol. 2008;15(3):683–688. doi: 10.1245/s10434-007-9747-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epi. 1992;45(6):613–619. doi: 10.1016/0895-4356(92)90133-8. [DOI] [PubMed] [Google Scholar]
- 21.Glatting G, Kletting P, Reske SN, Hohl K, Ring C. Choosing the optimal fit function: comparison of the Akaike information criterion and the F-test. Med Phys. 2007;34(100):4285–92. doi: 10.1118/1.2794176. [DOI] [PubMed] [Google Scholar]
- 22.Winkleby MA, Jatulis DE, Frank E, Fortmann SP. Socioeconomic status and health: how education, income, and occupation contribute to risk factors for cardiovascular disease. Am J Public Heath. 1992;82(6):816–820. doi: 10.2105/ajph.82.6.816. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Nishimura Y. Rationale for chemoradiotherapy. Int J Clin Oncol. 2004;9:414–420. doi: 10.1007/s10147-004-0443-z. [DOI] [PubMed] [Google Scholar]
- 24.Jones B, Dale RG. The potential for mathematical modeling in the assessment of the radiation dose equivalent of cytotoxic chemotherapy given concomitantly with radiotherapy. Br J Radiol. 2005;78:939–944. doi: 10.1259/bjr/40226390. [DOI] [PubMed] [Google Scholar]
- 25.Geh JI, Bond SJ, Bentzen SM, Glynne-Jones R. Systematic overview of preoperative (neoadjuvant) chemoradiotherapy trials in oesophageal cancer: evidence of a radiation and chemotherapy dose response. Radiother Oncol. 2006;78:236–244. doi: 10.1016/j.radonc.2006.01.009. [DOI] [PubMed] [Google Scholar]
- 26.Chen SW, Liang JA, Yang SN, Ko HL, Lin FJ. The adverse effect of treatment prolongation in cervical cancer by high-dose-rate intracavitary brachytherapy. Radiotherapy and Oncology. 2003;67:69–76. doi: 10.1016/s0167-8140(02)00439-5. [DOI] [PubMed] [Google Scholar]
- 27.Campbell SM, Roland MO, Buetow SA. Defining quality care. Social Science & Med. 2000;51:1611–1625. doi: 10.1016/s0277-9536(00)00057-5. [DOI] [PubMed] [Google Scholar]
- 28.Smith GL, Jiang J, Giordano SH, Meyer LA, Eifel PJ. Trends in the quality of treatment for patients with intact cervical cancer in the United States, 1999 through 2011. Int J Radiation Oncol Biol Phys. 2015;92(2):260–267. doi: 10.1016/j.ijrobp.2015.01.037. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lin JF, Berger JL, Krivak TC, Beriwal S, Chan JK, Sukumvanich P, Monk BJ, Richard SD. Impact of facility volume on therapy and survival for locally advanced cervical cancer. Gynecol Oncol. 2014;132:416–422. doi: 10.1016/j.ygyno.2013.12.013. [DOI] [PubMed] [Google Scholar]
- 30.American Society of Clinical Oncology. The state of cancer care in America, 2014: a report by the American Society of Clinical Oncology. J Clin Oncol. 2014;10(2):119–42. doi: 10.1200/JOP.2014.001386. [DOI] [PubMed] [Google Scholar]
- 31.Quality of Cancer Care Performance Measures. National Quality Forum website. [Accessed March 28, 2014]; http://www.qualityforum.org/Projects/c-d/Cancer_Care_Phase_II/Cancer_Care_Phase_I-II.aspx.
- 32.Lazar EJ, Fleischut P, Regan BK. Quality measurement in healthcare. Annu Rev Med. 2013;64:485–96. doi: 10.1146/annurev-med-061511-135544. [DOI] [PubMed] [Google Scholar]
- 33.Werner RM, Bradlow ET, Asch DA. Does hospital performance on process measures directly measure high quality care or is it just a marker of unmeasured care? Health Services Research. 2008;43(5):1464–1483. doi: 10.1111/j.1475-6773.2007.00817.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.