Abstract
Introduction:
Delays in radiation are multifactorial, frequent, and associated with poor outcomes. This study investigates the effect of both primary and adjuvant radiation therapy duration and their interaction with other measures of treatment delay on survival in head and neck squamous cell carcinoma (HNSCC).
Methods:
We built a retrospective cohort with the National Cancer Database, consisting of primary oral cavity, hypopharynx, larynx and oropharynx squamous cell carcinoma without distant metastasis with at least six weeks of radiation. The primary exposure was the duration of radiation therapy (DRT), and the primary outcome was death. We estimated the association between DRT and 5-year overall survival (OS) using Kaplan-Meier curves and hazard ratios (HRs) with Cox proportional hazard regression.
Results:
In both primary (definitive) and adjuvant (post-surgical) radiation settings, increased DRT results in decreased survival. In the primary radiation cohort, 5-year OS was 59.7% [59.1%–60.3%] among those with 47–53 days DRT, which decreased significantly with each subsequent week to completion (81+ days: 38.4% [36.2%–40.7%]). In the surgical cohort, survival decreased 16.5% when DRT took greater than 75+ days (40–46 days: 68.2% [67.3%–69.1%] vs. 75+ days: 53.3% [50.1%–56.7%]). Multivariate analyses showed increased hazard of death with increased DRT (primary radiation: 81+ days HR: 1.69 [1.58–1.81]); surgical: 75+ days HR: 1.61 [1.37–1.88]), with effects intensifying when restricting to those receiving full-dose radiation.
Conclusion:
A prolonged DRT was associated with worse OS in head and neck cancer. Radiation treatment delays of even a week lead to a significant survival disadvantage. DRT had a stronger association with survival than time to initiation of postoperative adjuvant radiotherapy.
Keywords: Head and neck cancer, radiation treatment, delays in treatment, Survival, Post-operative radiation therapy time
INTRODUCTION
Radiation therapy represents a therapeutic cornerstone in the management of head and neck squamous cell carcinoma (HNSCC) and is frequently used in both primary and adjuvant treatment settings. Prior research has established the importance of expedient treatment, identifying poorer outcomes in patients who experience delays during radiation therapy.[1–3] Delays in initiation of radiation therapy may be secondary to patient-related factors (e.g., socioeconomic barriers), physician-derived factors (e.g., delays in referrals or pre-treatment workup, dental evaluation), as well as systemic factors (e.g., restrictive prior authorizations).[4] Whereas, delays in duration of treatment can be attributed to factors related to toxicity management, staff and patient scheduling conflicts, and facility based issues such as machine maintenance.
In the primary setting, it is recommended that definitive therapy be initiated as soon as feasibly possible. Though it is impossible to define an acceptable time interval given the multifactorial nature of HNSCC and the heterogeneous patient population, it has been established that delays in treatment initiation are associated with increased tumor burden, worse locoregional control, and decreased overall survival in many patients.[5–7]
Adjuvant therapy, including postoperative radiation with or without chemotherapy, has been shown to improve locoregional control, and in many cases, overall survival in patients with advanced HNSCC.[8, 9] The National Cancer Center Network (NCCN) recommends postoperative radiation therapy (PORT) be initiated within six weeks of surgery for those with adverse pathologic features.[10] PORT started outside this time frame is associated with increased recurrence rates,[11, 12] with more recent studies suggesting delays may also influence survival outcomes.[1, 13] Additionally, overall “package time,” defined as the duration of time from surgery to completion of PORT, has been found to have implications for both locoregional control and survival.[1, 6, 14, 15] Recent literature suggests that package times of >11 weeks significantly increase recurrence and decrease survival, which may be related to accelerated tumor repopulation.[1, 2, 16]
While it is clear that delays in initiation of treatment and total package time affect outcomes, the independent effect of duration of radiation treatment (DRT) on survival remains unresolved. Here, we aim to build on previous knowledge to rigorously characterize the effect of DRT on HNSCC outcomes in both the primary and adjuvant setting.
METHODS
Study design
We constructed a retrospective cohort of head and neck cancer cases from 2010 to 2015 from the National Cancer Database (NCDB) – a joint program of the Commission on Cancer (CoC) of the American College of Surgeons and the American Cancer Society. To be included in this study (Figure 1), patients were restricted to first primary cancer of the oral cavity (ICO-O-3 codes: C02.0-C02.3, C02.8, C03.0-C03.9, C04.0-C04.9, C05.0, C05.2, C05.8, C06.0-C06.8), larynx (C32.0-C32.9), hypopharynx (C12.9-C12.9) or oropharynx (C10.0-C10.9; C0.51 C0.5.2; C01.9 or C02.4; C09.0 -C09.9) using codes defined from the Surveillance, Epidemiology, and End Results (SEER) Program. The pathology was restricted to squamous cell carcinoma (histology codes: 8052, 8083, 8078–8070). Patients must have codes for radiation and non-missing surgery codes. We also excluded cases receiving palliative care and cases with distant metastatic disease. Early glottic cancers (stage 1–2) were excluded due to differences in radiation dose and duration compared to most other head and neck cancers. Ultimately a total of 129,055 patients were available for further analyses (Figure 1). Due to different standard radiation completion times, we split these patients into two cohorts, a primary definitive radiation cohort of patients with DRT of at least 47 days (N=62,693) and a surgical cohort of patients with DRT of at least 40 days (N=38,520), based on standard fractionation practices for head and neck squamous cell carcinoma in the United States. [17–19]
Figure 1.
Flowchart of exclusion and inclusion and creation of primary radiation and surgical cohorts.
Exposure and Outcome assessment
Our primary exposure of interest is DRT (i.e., time from initiation to completion of radiation), which is a variable available in NCDB. We categorized DRT by the number of weeks to complete radiation therapy. We collapsed DRT times over 78 days in the primary radiation cohort and 71 days in the surgical cohort. Death was ascertained from NCDB for cases diagnosed before 2015. All cases were censored after five years of follow-up.
Covariates
We also adjusted for potential confounders in the association between DRT and survival. The covariates of interest included: insurance (uninsured, private insurance/managed care, Medicaid, Medicare, and other government insurance), site, HPV status, race, T category, N category, Charlson-Deyo comorbidity score (0 - no comorbidities, 1, and ≥2.), treatment, distance traveled to treatment and age.. The staging was based on the AJCC 7th edition staging guidelines. Pathologic stage was used when available, and clinical stage was used in its absence. HPV status was recorded under Collaborative State Site-Specific Factor 10.
Treatment initiation time was calculated as the time from diagnosis to the first course of treatment (i.e., either surgery or radiation) for both primary radiation and surgical cohorts. PORT time was calculated as the amount of time from the date of surgery to the initiation of radiotherapy. Additionally, given that the dosage of radiation treatment holds significant survival implications, we further categorized patients receiving radiation doses ≥ 66 Gy for the primary radiation cohort and ≥ 60 Gy for the surgical cohort, again based on conventional treatment protocols.7,16–18
Statistical Analyses
The descriptive analyses compared covariates across DRT for both primary and adjuvant radiation cohorts. We used a two-sided Pearson chi-square test for categorical variables. For continuous variables, means and standard deviations (SD) were calculated, and t-tests were conducted for normally distributed variables.
We calculated overall survival based on the time from the date of radiation completion to either date of death due to any cause, censoring due to loss of follow-up, or 5 years of follow-up. Kaplan-Meier all-cause survival plots were constructed, and log-rank p-values were calculated. Cox proportional hazards regression estimated hazard ratios (HR) for the independent effects of DRT on overall survival. We adjusted for the potential confounders described above. The proportional hazards assumption for DRT was tested and satisfied for all models. We also fit multivariate Cox regression models with a cubic natural spline analysis to determine the potential nonlinear association between DRT (continuous) and the risk of death. To keep outliers from influencing the plot, we only included cases who completed radiation within the 99th percentile (primary radiation: 100 days; surgical: 88 days). We used the rolr package in R to identify three-group splits based on a survival outcome. In short, rolr uses the hierarchical method in which it first identifies an optimal cutpoint that gives the largest logrank statistic to split into two groups, and then repeats the process in each of the resulting groupsto find additional two cutpoints. It then takes the cutpoint that gives the larger test statistic between the two as the second optimal cutpoint. We also examined the interaction between post-operative radiotherapy time and duration of radiation therapy. We also estimated restricted mean survival time (RMST), which is the area under the Kaplan-Meier curve and interpreted as the “life expectancy” between diagnosis and five years.[20] Alpha of 0.05 was used for all statistical testing and confidence interval calculations. Statistics including the survival analyses were conducted with R 3.2.0 using rms, survival, and survminer packages.
RESULTS
Descriptive Statistics
In the primary radiation cohort, our descriptive data revealed that minority patients and patients of low socioeconomic status were more likely to have DRT longer than 53 days (Table 1), with only 69.6% of all patients receiving radiation doses ≥66 Gy. Black Americans (Median: 55 [Interquartile range (IQR): 51–62]), females (Median: 54 [IQR: 51–61]), patients on Medicaid (Median: 55 [IQR: 51–62]) and patients treated in a community hospitals (Median: 55 [IQR: 51–62]) tended to have longer median DRT (Supplemental Table 1).
Table 1.
Descriptive statistics of the primary radiation cohort by duration of radiation therapy.
| Duration of Radiation Therapy(Days) | |||||||
|---|---|---|---|---|---|---|---|
| 47–53 | 54–60 | 61–67 | 68–73 | 74–80 | 81 + | ||
| N = 33346 | N = 16005 | N = 7051 | N = 2537 | N = 1496 | N = 2258 | ||
| N(%) | N(%) | N(%) | N(%) | N(%) | N(%) | p | |
| Sex | |||||||
| Male | 26889 (80.6) | 1236 8 (77.3) | 5352 (75.9) | 1838 (72.4) | 1072 (71.7) | 1655 (73.3) | <0.001 |
| Female | 6457 (19.4) | 3637 (22.7) | 1699 (24.1) | 699 (27.6) | 424 (28.3) | 603 (26.7) | |
| Insurance | |||||||
| Uninsured | 2026 (6.2) | 1157 (7.4) | 548 (7.9) | 208 (8.3) | 132 (9.0) | 203 (9.2) | <0.001 |
| Private | 15316 (46.6) | 6266 (39.8) | 2669 (38.5) | 881 (35.2) | 476 (32.5) | 693 (31.4) | |
| Medicaid | 3549 (10.8) | 2144 (13.6) | 1046 (15.1) | 440 (17.6) | 297 (20.3) | 432 (19.6) | |
| Medicare | 10979 (33.4) | 5729 (36.4) | 2488 (35.9) | 916 (36.6) | 528 (36.0) | 833 (37.7) | |
| Other government | 987 (3.0) | 438 (2.8) | 189 (2.7) | 55 (2.2) | 33 (2.3) | 48 (2.2) | |
| Missing | 489 | 271 | 111 | 37 | 30 | 49 | |
| Race | |||||||
| American Indian | 119 (0.4) | 53 (0.3) | 14 (0.2) | 12 (0.5) | 3 (0.2) | 8 (0.4) | <0.001 |
| Asian/Pacific Islander | 407 (1.2) | 147 (0.9) | 60 (0.9) | 21 (0.8) | 12 (0.8) | 24 (1.1) | |
| Black | 3424 (10.4) | 2145 (13.5) | 1060 (15.1) | 427 (17.0) | 232 (15.7) | 457 (20.4) | |
| Hispanic | 1052 (3.2) | 515 (3.3) | 279 (4.0) | 103 (4.1) | 74 (5.0) | 105 (4.7) | |
| White | 27982 (84.8) | 1297 7 (81.9) | 5592 (79.8) | 1956 (77.6) | 1157 (78.3) | 1645 (73.5) | |
| Charlson Comorbidity | |||||||
| 0 | 27008 (81.0) | 1263 9 (79.0) | 5528 (78.4) | 1972 (77.7) | 1138 (76.1) | 1734 (76.8) | <0.001 |
| 1 | 4946 (14.8) | 2536 (15.8) | 1177 (16.7) | 407 (16.0) | 281 (18.8) | 392 (17.4) | |
| 2+ | 1392 (4.2) | 830 (5.2) | 346 (4.9) | 158 (6.2) | 77 (5.1) | 132 (5.8) | |
| Facility Type | |||||||
| Community | 2836 (8.6) | 1853 (11.7) | 910 (13.1) | 361 (14.4) | 232 (15.7) | 337 (15.2) | <0.001 |
| Comprehensive Community | 13607 (41.3) | 6877 (43.5) | 2881 (41.5) | 1044 (41.5) | 574 (38.7) | 816 (36.8) | |
| Academic/Research | 12607 (38.2) | 5412 (34.2) | 2504 (36.0) | 877 (34.9) | 520 (35.1) | 816 (36.8) | |
| Integrated Network | 3935 (11.9) | 1678 (10.6) | 654 (9.4) | 232 (9.2) | 156 (10.5) | 246 (11.1) | |
| Missing | 361 | 185 | 102 | 23 | 14 | 43 | |
| Treatment | |||||||
| Radiation Only | 6111 (18.5) | 2726 (17.2) | 1037 (14.9) | 365 (14.5) | 213 (14.4) | 298 (13.3) | <0.001 |
| Radiation/Chemo | 26989 (81.5) | 1315 8 (82.8) | 5945 (85.1) | 2155 (85.5) | 1266 (85.6) | 1935 (86.7) | |
| Missing | 246 | 121 | 69 | 17 | 17 | 25 | |
| Site | |||||||
| Hypopharynx | 2719 (8.2) | 1471 (9.2) | 682 (9.7) | 247 (9.7) | 132 (8.8) | 205 (9.1) | <0.001 |
| Larynx | 9599 (28.8) | 4793 (29.9) | 2065 (29.3) | 773 (30.5) | 425 (28.4) | 651 (28.8) | |
| Oral Cavity | 1489 (4.5) | 923 (5.8) | 497 (7.0) | 170 (6.7) | 150 (10.0) | 219 (9.7) | |
| Oropharynx HPV-Negative | 2044 (6.1) | 823 (5.1) | 309 (4.4) | 96 (3.8) | 52 (3.5) | 93 (4.1) | |
| Oropharynx HPV-Positive | 3888 (11.7) | 1073 (6.7) | 392 (5.6) | 105 (4.1) | 66 (4.4) | 84 (3.7) | |
| Oropharynx Missing HPV | 13607 (40.8) | 6922 (43.2) | 3106 (44.1) | 1146 (45.2) | 671 (44.9) | 1006 (44.6) | |
| N Stage | |||||||
| N0 | 9011 (28.2) | 4453 (29.3) | 1811 (27.3) | 676 (28.1) | 400 (28.2) | 622 (29.4) | <0.001 |
| N1 | 5197 (16.3) | 2769 (18.2) | 1160 (17.5) | 402 (16.7) | 271 (19.1) | 375 (17.7) | |
| N2 | 16367 (51.3) | 7241 (47.6) | 3261 (49.2) | 1166 (48.5) | 647 (45.7) | 962 (45.5) | |
| N3 | 1347 (4.2) | 737 (4.8) | 393 (5.9) | 159 (6.6) | 98 (6.9) | 154 (7.3) | |
| Missing | 1424 | 805 | 426 | 134 | 80 | 145 | |
| T Stage | |||||||
| T1 | 5267 (16.7) | 2142 (14.3) | 807 (12.3) | 255 (10.7) | 152 (10.8) | 207 (9.9) | <0.001 |
| T2 | 11760 (37.3) | 5426 (36.1) | 2323 (35.4) | 836 (35.2) | 443 (31.6) | 651 (31.1) | |
| T3 | 8967 (28.5) | 4528 (30.1) | 1981 (30.2) | 716 (30.2) | 463 (33.0) | 721 (34.4) | |
| T4 | 5506 (17.5) | 2934 (19.5) | 1455 (22.2) | 567 (23.9) | 345 (24.6) | 514 (24.6) | |
| Missing | 1846 | 975 | 485 | 163 | 93 | 165 | |
| Radiation > 66 gy | |||||||
| No | 9951 (30.4) | 6694 (42.8) | 3172 (47.2) | 1167 (48.3) | 689 (49.1) | 965 (48.3) | <0.001 |
| Yes | 22736 (69.6) | 8942 (57.2) | 3542 (52.8) | 1251 (51.7) | 714 (50.9) | 1034 (51.7) | |
| Missing | 656 | 369 | 337 | 119 | 93 | 259 | |
| Time to Treatment Initiation (Days) | |||||||
| <28 | 6908 (20.7) | 3482 (21.8) | 1533 (21.7) | 511 (20.1) | 324 (21.7) | 603 (26.7) | <0.001 |
| 28 to 56 | 15758 (47.3) | 7043 (44.0) | 2814 (39.9) | 1048 (41.3) | 602 (40.2) | 876 (38.8) | |
| 56+ | 10680 (32.0) | 5480 (34.2) | 2704 (38.3) | 978 (38.5) | 570 (38.1) | 779 (34.5) | |
| Distance traveled (mean(SD)) | 26.84 (104.67) | 21.66 (71.21) | 22.40 (80.93) | 22.15 (73.54) | 18.47 (38.09) | 20.24 (66.40) | <0.001 |
| Age (mean(SD)) | 60.46 (10.08) | 60.63 (10.43) | 60.27 (10.61) | 60.39 (10.13) | 60.41 (10.98) | 59.97 (10.73) | 0.035 |
IQR: Interquartile Range; SD: Standard deviation
We saw similar results in the surgical cohort (Table 2), with only 67.7% of patients receiving radiation doses ≥60 Gy. Similar to the definitive radiation cohort, Black Americans (Median: 50 [IQR: 45.0–56.0]), patients treated in a community hospital (Median: 51 [IQR: 47–58]), those who had trimodality treatment (Median: 50 [IQR: 46–56]), and those with positive margins (Median: 50 [IQR: 46–55]) were more likely to have prolonged DRT (Supplemental Table 1).
Table 2.
Descriptive statistics of the surgical cohort by duration of radiation therapy.
| Duration of Radiation Therapy(Days) | |||||||
|---|---|---|---|---|---|---|---|
| 40–46 | 47–53 | 54–60 | 61–67 | 68–74 | 75+ | ||
| N = 13868 | N = 14773 | N = 5489 | N = 2403 | N = 900 | N = 1087 | ||
| N(%) | N(%) | N(%) | N(%) | N(%) | N(%) | p | |
| Sex | |||||||
| Male | 10569 (76.2) | 11398 (77.2) | 4147 (75.6) | 1790 (74.5) | 647 (71.9) | 785 (72.2) | <0.001 |
| Female | 3299 (23.8) | 3375 (22.8) | 1342 (24.4) | 613 (25.5) | 253 (28.1) | 302 (27.8) | |
| Insurance | |||||||
| Uninsured | 650 (4.8) | 755 (5.2) | 336 (6.2) | 136 (5.7) | 73 (8.2) | 66 (6.2) | <0.001 |
| Private | 7452 (54.5) | 7829 (53.7) | 2694 (49.7) | 1128 (47.6) | 389 (43.9) | 431 (40.5) | |
| Medicaid | 1301 (9.5) | 1408 (9.7) | 625 (11.5) | 302 (12.7) | 125 (14.1) | 190 (17.9) | |
| Medicare | 4000 (29.2) | 4249 (29.1) | 1644 (30.3) | 759 (32.0) | 286 (32.3) | 357 (33.6) | |
| Other government | 280 (2.0) | 342 (2.3) | 122 (2.3) | 44 (1.9) | 13 (1.5) | 20 (1.9) | |
| Missing | 185 | 190 | 68 | 34 | 14 | 23 | |
| Race | |||||||
| American Indian | 31 (0.2) | 43 (0.3) | 15 (0.3) | 7 (0.3) | 0 (0.0) | 3 (0.3) | <0.001 |
| Asian/Pacific Islander | 279 (2.0) | 255 (1.7) | 78 (1.4) | 29 (1.2) | 16 (1.8) | 10 (0.9) | |
| Black | 1021 (7.4) | 1207 (8.3) | 510 (9.4) | 238 (10.0) | 115 (13.0) | 138 (12.9) | |
| Hispanic | 465 (3.4) | 487 (3.3) | 209 (3.8) | 95 (4.0) | 41 (4.6) | 68 (6.4) | |
| White | 11914 (86.9) | 12606 (86.4) | 4617 (85.0) | 2012 (84.5) | 713 (80.6) | 851 (79.5) | |
| Charlson Comorbidity | |||||||
| 0 | 11059 (79.7) | 11862 (80.3) | 4344 (79.1) | 1866 (77.7) | 679 (75.4) | 820 (75.4) | <0.001 |
| 1 | 2257 (16.3) | 2341 (15.8) | 899 (16.4) | 417 (17.4) | 170 (18.9) | 211 (19.4) | |
| 2+ | 552 (4.0) | 570 (3.9) | 246 (4.5) | 120 (5.0) | 51 (5.7) | 56 (5.2) | |
| Facility Type | |||||||
| Community | 747 (5.6) | 1158 (8.1) | 579 (10.9) | 258 (11.1) | 132 (15.3) | 133 (12.7) | <0.001 |
| Comprehensive Community | 4180 (31.1) | 5689 (39.6) | 2167 (40.6) | 898 (38.6) | 298 (34.5) | 377 (36.0) | |
| Academic/Research | 7074 (52.6) | 6025 (41.9) | 2056 (38.6) | 961 (41.3) | 351 (40.7) | 439 (41.9) | |
| Integrated Network | 1446 (10.8) | 1502 (10.4) | 531 (10.0) | 211 (9.1) | 82 (9.5) | 98 (9.4) | |
| Missing | 421 | 399 | 156 | 75 | 37 | 40 | |
| Treatment | |||||||
| Surgery/Radiation | 7192 (52.8) | 5508 (37.9) | 1841 (34.0) | 748 (31.7) | 244 (27.6) | 355 (33.4) | <0.001 |
| Surgery/Radiation/Chemo | 6428 (47.2) | 9043 (62.1) | 3575 (66.0) | 1614 (68.3) | 641 (72.4) | 707 (66.6) | |
| Missing | 248 | 222 | 73 | 41 | 15 | 25 | |
| Site | |||||||
| Hypopharynx | 390 (2.8) | 457 (3.1) | 203 (3.7) | 73 (3.0) | 38 (4.2) | 34 (3.1) | <0.001 |
| Larynx | 2570 (18.5) | 2735 (18.5) | 1022 (18.6) | 465 (19.4) | 193 (21.4) | 222 (20.4) | |
| Oral Cavity | 4812 (34.7) | 3752 (25.4) | 1293 (23.6) | 633 (26.3) | 236 (26.2) | 336 (30.9) | |
| Oropharynx HPV-Negative | 563 (4.1) | 674 (4.6) | 210 (3.8) | 80 (3.3) | 24 (2.7) | 35 (3.2) | |
| Oropharynx HPV-Positive | 1763 (12.7) | 1751 (11.9) | 471 (8.6) | 177 (7.4) | 56 (6.2) | 66 (6.1) | |
| Oropharynx Missing HPV | 3770 (27.2) | 5404 (36.6) | 2290 (41.7) | 975 (40.6) | 353 (39.2) | 394 (36.2) | |
| N Stage | |||||||
| N0 | 4584 (40.2) | 4135 (34.4) | 1458 (33.5) | 652 (33.9) | 221 (31.0) | 291 (35.2) | <0.001 |
| N1 | 2257 (19.8) | 2435 (20.2) | 889 (20.5) | 368 (19.1) | 150 (21.1) | 151 (18.3) | |
| N2 | 4395 (38.5) | 5190 (43.1) | 1867 (43.0) | 830 (43.1) | 320 (44.9) | 362 (43.8) | |
| N3 | 181 (1.6) | 274 (2.3) | 132 (3.0) | 76 (3.9) | 21 (2.9) | 22 (2.7) | |
| Missing | 2451 | 2739 | 1143 | 477 | 188 | 261 | |
| T Stage | |||||||
| T1 | 2838 (26.9) | 3011 (27.0) | 1016 (24.8) | 418 (23.0) | 151 (22.7) | 154 (19.7) | <0.001 |
| T2 | 3775 (35.7) | 4207 (37.7) | 1529 (37.3) | 672 (37.0) | 228 (34.3) | 276 (35.4) | |
| T3 | 1521 (14.4) | 1755 (15.7) | 742 (18.1) | 346 (19.1) | 135 (20.3) | 149 (19.1) | |
| T4 | 2433 (23.0) | 2184 (19.6) | 817 (19.9) | 380 (20.9) | 151 (22.7) | 201 (25.8) | |
| Missing | 3301 | 3616 | 1385 | 587 | 235 | 307 | |
| Margin Status | |||||||
| Negative | 10036 (85.6) | 8117 (75.5) | 2871 (74.8) | 1298 (75.4) | 477 (79.0) | 592 (78.6) | <0.001 |
| Positive | 1684 (14.4) | 2635 (24.5) | 965 (25.2) | 423 (24.6) | 127 (21.0) | 161 (21.4) | |
| Missing | 2148 | 4021 | 1653 | 682 | 296 | 334 | |
| Radiation ≥ 60 gy | |||||||
| No | 3597 (26.7) | 4521 (31.4) | 2170 (41.0) | 944 (43.2) | 359 (43.3) | 387 (43.2) | <0.001 |
| Yes | 9889 (73.3) | 9856 (68.6) | 3118 (59.0) | 1240 (56.8) | 470 (56.7) | 509 (56.8) | |
| Missing | 382 | 396 | 201 | 219 | 71 | 191 | |
| Time to Treatment Initiation (Days) | |||||||
| <28 | 7817 (56.4) | 9865 (66.8) | 3711 (67.6) | 1586 (66.0) | 611 (67.9) | 705 (64.9) | <0.001 |
| 28 to 56 | 4458 (32.1) | 3604 (24.4) | 1287 (23.4) | 581 (24.2) | 200 (22.2) | 254 (23.4) | |
| 56+ | 1593 (11.5) | 1304 (8.8) | 491 (8.9) | 236 (9.8) | 89 (9.9) | 128 (11.8) | |
| Post-Operative Radiation Time (Days) | |||||||
| <42 | 5094 (36.7) | 6311 (42.7) | 2345 (42.7) | 1008 (41.9) | 375 (41.7) | 488 (44.9) | <0.001 |
| 42 to 48 | 2940 (21.2) | 2767 (18.7) | 895 (16.3) | 357 (14.9) | 132 (14.7) | 151 (13.9) | |
| 49+ | 5834 (42.1) | 5695 (38.6) | 2249 (41.0) | 1038 (43.2) | 393 (43.7) | 448 (41.2) | |
| Distance traveled (mean(SD)) | 40.02 (132.25) | 29.73 (110.03) | 27.46 (92.45) | 28.40 (57.59) | 29.56 (77.65) | 29.65 (74.11) | <0.001 |
| Age (mean(SD)) | 58.93 (10.90) | 58.57 (10.67) | 58.54 (10.65) | 58.55 (10.88) | 58.30 (11.37) | 58.59 (11.13) | 0.044 |
IQR: Interquartile Range; SD: Standard deviation
Univariate Survival
Kaplan-Meier curves (Figure 2a) demonstrated significant differences in five-year overall survival (OS) by DRT (p-value < 0.001) in the primary radiation cohort. Five-year OS was the highest among those with DRT of 47 to 53 days (59.7% [95% Confidence Interval (CI): 59.1%–60.3%]), with decreased survival with each additional week of DRT. When we restricted the cohort to cases who had at least 66 Gy, we saw nearly identical results (Supplemental Figure 1a).
Figure 2.
Kaplan Meier curves by duration of radiation therapy for all cases in the A) primary radiation cohort and B) surgical cohort.
For the surgical cohort, the trends were similar to those seen in the primary radiation cohort, but survival overall was higher (Figure 2b). There was no difference in five-year OS between patients with DRT of 40–46 days (68.2% [95%CI: 67.3%–69.1%)]) versus 47–53 days (67.3% [95%CI: 66.5%–68.2%]). However, five-year OS again decreased significantly with DRT of 54–60 days and 61–67 days, while DRT of 75+ days was not significantly different from 68–74 days. When restricted to patients who had at least 60 Gy, survival as a whole improved, although the negative impact of prolonged DRT was maintained (Supplemental Figure 1b).
Multivariable Survival
In the primary radiation cohort, when adjusted for potential confounders, we found the risk of death increased with prolonged DRT (Table 3). Patients with DRT longer than 81+ days had significantly worse survival (HR: 1.69 [95%CI: 1.58–1.81]) compared to those who completed radiation in 47–53 days. When restricted to cases that received at least 66 Gy, the HRs become even more pronounced (Supplemental table 2). When plotting the HR against DRT in an adjusted Cox regression model, an S-shaped trend was seen, where the HR consistently increases after 50 days followed by a diminutive effect after 70 days (Figure 3a). This curve demonstrated a HR that crossed 1.0 (i.e., increased risk of death) at 53 days, suggesting that treatment delays extending DRT beyond this point may meaningfully influence survival outcomes.
Table 3.
Mutually-adjusted hazard ratios of all cases in the primary radiation cohort and surgical cohort.
| Primary Radiation | Surgical Cohort | ||||
|---|---|---|---|---|---|
| HR(95% CI) | p-value | HR*(95 % CI) | p-value | ||
| Duration of Radiation Therapy (Days) | Duration of Radiation Therapy (Days) | ||||
| 47–53 | 1 | 40–46 | 1 | ||
| 54–60 | 1.23 (1.19, 1.27) | <0.001 | 47–53 | 1.11 (1.04, 1.18) | 0.001 |
| 61–67 | 1.33 (1.28, 1.39) | <0.001 | 54–60 | 1.34 (1.24, 1.46) | <0.001 |
| 68–73 | 1.5 (1.41, 1.6) | <0.001 | 61–67 | 1.47 (1.31, 1.63) | <0.001 |
| 74–80 | 1.65 (1.53, 1.78) | <0.001 | 68–74 | 1.34 (1.13, 1.59) | 0.001 |
| 81 + | 1.69 (1.58, 1.81) | <0.001 | 75+ | 1.61 (1.37, 1.88) | <0.001 |
| Time to Treatment Initiation (Days) | Time to Treatment Initiation (Days) | ||||
| <28 | 1 | <28 | 1 | ||
| 28 to 56 | 0.98 (0.94, 1.01) | 0.231 | 28 to 56 | 1.04 (0.98, 1.1) | 0.199 |
| 56+ | 1.05 (1.01, 1.09) | 0.017 | 56+ | 1.01 (0.93, 1.1) | 0.78 |
| Radiation Dose | Radiation Dose | ||||
| <66 gy | 1 | <60 gy | 1 | ||
| ≥66 gy | 0.91 (0.88, 0.93) | <0.001 | ≥60 gy | 0.96 (0.91, 1.02) | 0.162 |
| Post-Operative Adjuvant Radiation Time (Days) | |||||
| -- | -- | -- | <42 | 1 | |
| -- | -- | -- | 42 to 48 | 1.08 (1.00, 1.17) | 0.044 |
| -- | -- | -- | 49+ | 1.19 (1.12, 1.27) | <0.001 |
| -- | -- | -- | Margin Status | ||
| -- | -- | -- | Negative | 1 | |
| -- | -- | -- | Positive | 1.21 (1.14, 1.3) | <0.001 |
Adjusted for insurance, median income quartile, site, race, T stage, N stage, charlson comorbidity, treatment, age
HR: Hazard Ratio; CI: Confidence Interval
Figure 3.
Adjusted Cox regression models with a cubic natural spline for duration of radiation therapy among the A) primary radiation and B) surgical cohort.
When adjusted for potential confounders in the surgical cohort, we saw similar trends as with primary radiation, even after adjusting for other treatment delays including delays in surgery as well as delays in initiation of PORT (Table 3). Patients with DRT longer than 75 days had worse survival (HR: 1.61 [95% CI: 1.37–1.88]) compared to those with DRT of 40–46 days. When restricted to cases who received more than 60 Gy, the HRs became stronger (Supplemental table 2). Similar to the primary radiation cohort, when plotting HR against DRT, the HR consistently increases after 50 days, followed by a diminutive effect around 60 days in an adjusted Cox regression model (Figure 3). This curve again demonstrated an HR with increased risk of death (HR >1.0) around 50 days, suggesting that DRT prolonged beyond this point leads to poorer survival outcomes.
Cut Point Analysis
With the rolr package in R, in the primary radiation cohort, we found two cut points at 54 days and 69 days (Supplemental Figure 2a). Patients with DRT <54 days had a five-year OS of 59.7% (95%CI: 59.3%–60.3%), which was higher than patients with DRT of 54 to 69 days (five-year OS: 49.8%% [95%CI: 49.1%–50.5%]). After adjusting for potential confounders, patients with DRT >70 days had 1.65 (95%CI: 1.58–1.72) times the hazard of death compared to those who completed radiation in less than 54 days.
In the surgical cohort, we found similar cut points at 55 and 64 days (Supplemental Figure 2b). Patients with DRT <55 days had a five-year OS of 67.6% (95%CI: 67.01%–68.2%), while patients with DRT >64 days had a five-year OS of 62.4% (95%CI: 61.1%–63.8%). After adjusting for potential confounders, compared with patients with DRT <55 days, patients who completed radiation in >64 days had a HR of 1.47 (95%CI: 1.34–1.60).
Relative Impact of Delays in PORT Initiation Versus Duration in the Surgical Cohort
Although there was no interaction between time to PORT initiation and DRT, we found that DRT had a stronger, independent association with survival. Among patients with DRT <55 days, delays in PORT initiation from 42 to 48 days increased the hazard of death by 7.0%, while delays in PORT initiation 49 days and beyond increased the hazard of death by 17% relative to patients with no delays in PORT initiation (i.e., <42 days) (Table 4). Strikingly, among patients with no delay in PORT initiation, patients with DRT between 55 and 64 days had 1.21 times the hazard of death, and those with DRT >64 days had 1.39 times the hazard of death compared to patients who completed radiation in <55 days.
Table 4.
Adjusted Hazard Ratio and unadjusted Restricted Mean Survival Time for post-operative adjuvant radiotherapy initiation time and duration of radiation therapy.
| PORT Initiation Time(Days) | Duration of Radiation Therapy(Days) | HR(95% CI) | RMST(95% CI) |
|---|---|---|---|
| <42 | <55 | 1 | 4.31 (4.28, 4.33) |
| <42 | 55–64 | 1.21 (1.07, 1.37) | 4.08 (4.02, 4.14) |
| <42 | 64+ | 1.39 (1.20, 1.62) | 3.82 (3.73, 3.91) |
| 42 to 48 | <55 | 1.07 (0.99, 1.17) | 4.10 (4.06, 4.14) |
| 42 to 48 | 55–64 | 1.36 (1.14, 1.61) | 4.00 (3.90, 4.11) |
| 42 to 48 | 64+ | 1.40 (1.13, 1.74) | 3.71 (3.55, 3.87) |
| 49+ | <55 | 1.17 (1.09, 1.25) | 3.95 (3.92, 3.98) |
| 49+ | 55–64 | 1.49 (1.35, 1.66) | 3.71 (3.64, 3.77) |
| 49+ | 64+ | 1.84 (1.62, 2.09) | 3.47 (3.38, 3.57) |
PORT: Post-Operative Adjuvant Radiation Time; HR: Hazard Ratio; CI: Confidence Interval; RMST: Restricted Mean Survival Time
These results were also reflected in restricted mean survival time (RMST) analyses (Table 4). The average survival time from completion of radiation bounded at five years was 4.31 (95%CI: 4.28, 4.33) years in patients with no delays. However, for patients with up to a 1-week delay in PORT initiation but no delays in DRT, the RMST was 4.10 (95%CI: 4.06, 4.14). In contrast, patients with up to a 9-day delay in DRT and no delay in PORT initiation had a slightly worse RMST of 4.08 (95%CI: 4.02, 4.14). Thus, by both HR as well as RMST analyses, delays in DRT appear to have a larger effect on survival than delays in initiation of PORT.
DISCUSSION
In this large retrospective national cohort of head and neck cancer patients, we found prolonged DRT to be associated with an increased hazard of death, even after adjusting for time to initiation of radiotherapy as well as completion of a full standard dose of radiotherapy, in both the primary and adjuvant (postoperative) treatment settings. While studies have evaluated the effects of delays in the initiation of radiation and total package time, there is an absence of rigorous analyses of DRT and its impact on outcomes in HNSCC. Here, we sought to specifically explore this portion of a patient’s treatment, focusing on head and neck patients who underwent surgery with adjuvant therapy as well as definitive primary radiation therapy, while simultaneously adjusting for the significant confounding effects of HPV status and subsite.
Strikingly, we saw a significant decrease in survival associated with each additional week of DRT in both the primary and adjuvant setting – an effect which was augmented when further restricting to those that received full radiation doses. Increased DRT appeared to be most profoundly associated with poor survival in the primary radiation cohort, where overall survival decreased by nearly 22% when DRT was extended beyond 81 days. These results suggest timely completion of radiation, without delay, is essential for optimizing patient outcomes. More specifically, our data indicate that providing patients a treatment break to allow toxicities to resolve should be strongly avoided and only undertaken after acknowledging a detriment in oncologic outcomes. In addition, in cases where delays in treatment are unavoidable, consideration should be given to accelerating the remaining treatment course to minimize DRT.
In order to contextualize the significance of prolonged DRT, it can be compared to other well-known survival determinants. In the primary radiation cohort, among those receiving a full dose of radiation, the HR for just a single week of prolonged DRT (HR 1.29 [95%CI: 1.24–1.35]) is greater than delaying initiation of therapy by an entire month (HR 1.02 [95% CI: 0.97–1.07]). Within the surgical cohort among those who received a full dose of PORT, a two-week delay in DRT (HR 1.42 [95%CI: 1.29–1.58]) had a relatively greater impact on survival than delays in initiation of PORT beyond 7 weeks (HR 1.18 [95%CI: 1.09–1.27]). Furthermore, comparing these numbers to well-described metrics in the surgical literature is even more humbling: A two-week delay in DRT correlated with a greater hazard of death than even positive margins (HR 1.23 [95%CI: 1.14–1.34]).
Therefore, it is of significant concern that 47% of the primary radiation group, and 64% of the surgical group failed to complete radiation within the 47–53 and 40–46 day recommended time frames, respectively. Given that we saw no significant decline in survival associated with a one week delay in DRT in the surgical group, it is, however, encouraging 74% of patients completed adjuvant radiation therapy within 53 days of starting treatment. Together, these data suggest that there may be a small degree of flexibility in DRT for the surgical cohort, with a uniform target of 7 weeks (+/− 2 days) for completion of primary or adjuvant radiation therapy, perhaps representing a reasonable oncologic goal.
In comparing the surgical and primary radiation cohorts, DRT delays were seen more commonly in the surgical group. This is interesting given that, for surgically treated patients, DRT delays were more common than delays in initiation of adjuvant therapy. Over 36% of patients initiated adjuvant therapy within the recommended 6-week postoperative window. Contrary to commonly held assumptions, delays in PORT initiation did not increase the likelihood of delays in DRT (data not shown). While we are unable to determine the cause of lengthened DRT among this cohort, post-surgical patients may be more susceptible to treatment complications or delays related to wound healing. Also, radiation in the adjuvant setting may prove more mentally, physically, emotionally, or socioeconomically tolling (after prior surgery), compared to patients who are receiving primary radiation therapy.
Importantly, our findings differ somewhat from previous analyses on the impact of delays in adjuvant therapy on survival in head and neck cancer. A study by Ho et al. evaluated the impact of DRT delays in patients with head and neck cancer undergoing surgery and PORT.[15] They found that each additional day of delay in radiation duration was associated with continuous, escalated mortality up to 55 days (HR 1.25, 95%CI: 1.11, 1.41). However, in contrast to our findings demonstrating a continued, albeit diminutive, increase in the HR with lengthening DRT, they identified no detrimental impact on survival with delays beyond 55 days.13 These differences are likely attributable to important variations between study populations, such as an exclusively surgical cohort from 2004–2013 and exclusion of patients who received less than 60 Gy or those receiving chemotherapy > 14 days before or after initiating radiation, as well as methodology. Although a subgroup analysis of HPV-positive oropharyngeal cancer patients was completed, they did not control for HPV status in multivariable analyses, which represents a significant confounder in the analysis of their outcomes. Importantly, our study extends this crucial area of research on DRT in new directions. We included a temporally matched, primary radiation cohort, where we found that prolonged DRT had the most significant impact on mortality compared to other modifiable factors such as time to initiation of treatment. By addressing the limitations of prior approaches and investigating both surgical and primary radiation cohorts, our analyses offer more reliable and generalizable conclusions across a broad cross-section of head and neck cancer patients.
Our findings draw attention to important considerations regarding the management and counseling of patients requiring both primary (definitive) and adjuvant radiation therapy. Guidelines regarding PORT have overwhelmingly focused on minimizing the time between surgery and initiation of adjuvant therapy; however, our study fortifies the point that DRT is more consequential as it relates to survival outcomes. Importantly, although delays in initiating PORT are associated with an increased hazard of death, the increase in the hazard of death is smaller than that associated with delays in DRT. Thus, our results suggest that if a patient needs additional recovery or healing time, delaying the start of PORT may be more favorable than delaying radiation completion once it has begun.
Notably, controlling for DRT can be challenging given the multitude of associated factors. In our study, we found that delays were more common among minority patients and those with lower socioeconomic status, which suggests that the DRT likely contributes to the survival disparities associated with socioeconomic status and race in head and neck cancer. Additionally, circumstances related to post-surgical healing, treatment-related side effects, and patient performance status may also influence DRT. With this in mind, multidisciplinary care teams must work together on an individual and systemic level to mitigate any actionable variables that may contribute to delays in DRT in both the definitive and adjuvant radiation therapy settings.
We acknowledge several limitations associated with this study. First, it is important to highlight the context in which we interpreted our results as related to common national radiation/chemotherapy practices. Standard definitive radiation fractionation regimens for HNSCC in the United States consist of 66–72 Gy delivered over approximately 7 weeks in once-daily fractions of 1.8 to 2.0 Gy given 5 days per week. Conventional adjuvant PORT regimens deliver 60 Gy in 2.0 Gy once-daily increments 5 days a week over a 6 week period postoperatively.[17–19] These should be acknowledged as a general representation of common practice, with the caveat that several prospective, randomized controlled trials have demonstrated the utility of both accelerated and hyperfractionated regimens in specific clinical scenarios. The inability to control for such alternate treatment regimens in the current study represents a limitation. However, we anticipate that the vast majority of treatment regimens administered are similar, if not identical, to standard protocols, which may mitigate the impact of heterogeneity in radiation protocols on our findings. In addition, although NCDB is a prospectively collected database, this study is retrospective, potentially enabling unmeasured variables and misclassification errors, including misclassification of DRT. We attempted to mitigate this possibility by only including patients with plausible radiation treatment doses. Additionally, we do not have detailed clinical and demographic information on these patients including prescribed dose versus completion of treatment, reason for incomplete treatment, and smoking data. Smoking is a potential confounder. However, the relationship between smoking and DRT is likely weak, which suggests smoking does not confound this relationship completely. NCDB does, however, offer more detailed treatment data than SEER, and we believe our study offers a valuable depiction of the survival implications associated with delays in DRT for both primary and adjuvant radiation therapy.
In conclusion, we found that delays in DRT are strongly associated with poor survival in head and neck cancer among patients undergoing primary radiation as well as postoperative adjuvant radiation. Delays in DRT of even a week appear to be associated with significantly reduced survival. Notably, the effect of prolonged DRT on survival is more substantial than delays in treatment initiation. Collectively, our findings highlight the importance of radiation therapy adherence and avoiding treatment breaks to allow toxicity to resolve, emphasizing an important role for future research into identifying factors that trigger delays in DRT and leveraging cut points outlined here to inform quality improvement guidelines.
Supplementary Material
Research Highlights.
Delays in radiation treatment led to worse overall survival in HNSCC patients.
A large proportion of the cohort does not complete radiation within guidelines.
Delays in duration of radiation treatment is more predictive than prolonged times to initiation of post-operative radiotherapy.
Funding:
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
There are no conflicts of interest to disclose.
REFERENCES
- [1].Chen MM, Harris JP, Orosco RK, Sirjani D, Hara W, Divi V. Association of Time between Surgery and Adjuvant Therapy with Survival in Oral Cavity Cancer. Otolaryngol Head Neck Surg. 2018;158:1051–6. [DOI] [PubMed] [Google Scholar]
- [2].Graboyes EM, Kompelli AR, Neskey DM, Brennan E, Nguyen S, Sterba KR, et al. Association of Treatment Delays With Survival for Patients With Head and Neck Cancer: A Systematic Review. JAMA Otolaryngol Head Neck Surg. 2019;145:166–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Shaikh T, Handorf EA, Murphy CT, Mehra R, Ridge JA, Galloway TJ. The Impact of Radiation Treatment Time on Survival in Patients With Head and Neck Cancer. Int J Radiat Oncol Biol Phys. 2016;96:967–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Neal RD. Do diagnostic delays in cancer matter? Br J Cancer. 2009;101 Suppl 2:S9–S12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Coca-Pelaz A, Takes RP, Hutcheson K, Saba NF, Haigentz M Jr., Bradford CR, et al. Head and Neck Cancer: A Review of the Impact of Treatment Delay on Outcome. Advances in therapy. 2018;35:153–60. [DOI] [PubMed] [Google Scholar]
- [6].Chen Z, King W, Pearcey R, Kerba M, Mackillop WJ. The relationship between waiting time for radiotherapy and clinical outcomes: a systematic review of the literature. Radiother Oncol. 2008;87:3–16. [DOI] [PubMed] [Google Scholar]
- [7].Fortin A, Bairati I, Albert M, Moore L, Allard J, Couture C. Effect of treatment delay on outcome of patients with early-stage head-and-neck carcinoma receiving radical radiotherapy. Int J Radiat Oncol Biol Phys. 2002;52:929–36. [DOI] [PubMed] [Google Scholar]
- [8].Bernier J, Domenge C, Ozsahin M, Matuszewska K, Lefebvre JL, Greiner RH, et al. Postoperative irradiation with or without concomitant chemotherapy for locally advanced head and neck cancer. The New England journal of medicine. 2004;350:1945–52. [DOI] [PubMed] [Google Scholar]
- [9].Cooper JS, Zhang Q, Pajak TF, Forastiere AA, Jacobs J, Saxman SB, et al. Long-term follow-up of the RTOG 9501/intergroup phase III trial: postoperative concurrent radiation therapy and chemotherapy in high-risk squamous cell carcinoma of the head and neck. Int J Radiat Oncol Biol Phys. 2012;84:1198–205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].National Comprehensive Cancer Network. National Comprehensive Cancer Network head and neck cancers. 2018.
- [11].Ang KK, Trotti A, Brown BW, Garden AS, Foote RL, Morrison WH, et al. Randomized trial addressing risk features and time factors of surgery plus radiotherapy in advanced head-and-neck cancer. Cancer/Radiothérapie. 2002;6:259–60. [DOI] [PubMed] [Google Scholar]
- [12].Huang J, Barbera L, Brouwers M, Browman G, Mackillop WJ. Does delay in starting treatment affect the outcomes of radiotherapy? A systematic review. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2003;21:555–63. [DOI] [PubMed] [Google Scholar]
- [13].Graboyes EM, Garrett-Mayer E, Ellis MA, Sharma AK, Wahlquist AE, Lentsch EJ, et al. Effect of time to initiation of postoperative radiation therapy on survival in surgically managed head and neck cancer. Cancer. 2017;123:4841–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Harris JP, Chen MM, Divi V, Hara WY. Shorter Package Times Improve Survival for Head and Neck Cancer Patients in the United States. International Journal of Radiation Oncology*Biology*Physics. 2016;96:E381–E2. [Google Scholar]
- [15].Ho AS, Kim S, Tighiouart M, Mita A, Scher KS, Epstein JB, et al. Quantitative survival impact of composite treatment delays in head and neck cancer. Cancer. 2018;124:3154–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Tubiana M. Repopulation in human tumors. A biological background for fractionation in radiotherapy. Acta Oncol. 1988;27:83–8. [DOI] [PubMed] [Google Scholar]
- [17].Corvo R. Evidence-based radiation oncology in head and neck squamous cell carcinoma. Radiother Oncol. 2007;85:156–70. [DOI] [PubMed] [Google Scholar]
- [18].Rosenthal DI, Blanco AI. Head and neck squamous cell carcinoma: optimizing the therapeutic index. Expert Rev Anticancer Ther. 2005;5:501–14. [DOI] [PubMed] [Google Scholar]
- [19].Lacas B, Bourhis J, Overgaard J, Zhang Q, Gregoire V, Nankivell M, et al. Role of radiotherapy fractionation in head and neck cancers (MARCH): an updated meta-analysis. The Lancet Oncology. 2017;18:1221–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Royston P, Parmar MK. Restricted mean survival time: an alternative to the hazard ratio for the design and analysis of randomized trials with a time-to-event outcome. BMC Med Res Methodol. 2013;13:152. [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.