Quantitative survival impact of composite treatment delays in head and neck cancer

Abstract

Background

Multidisciplinary head and neck cancer (HNC) management must reconcile increasingly sophisticated subspecialty care with timeliness of care. Prior studies have examined the individual effects of delays in diagnosis to treatment interval (DTI), postoperative interval, and radiation interval but not considered them collectively. We investigate the combined impact of these interwoven intervals on HNC patients.

Methods

HNC patients undergoing curative-intent surgery with radiation were identified in the National Cancer Database between 2004-2013. Multivariable models were constructed using restricted cubic splines to determine non-linear relationships with overall survival.

Results

Overall, 15,064 patients were evaluated. After adjustment for covariates, only prolonged postoperative interval (p<0.001) and radiation intervals (p<0.001) independently predicted for worse outcomes, while the association of DTI with survival disappeared. Using multivariable restricted cubic spline functions, increasing postoperative interval did not affect mortality until 40 days after surgery, with each day of delay beyond this increasing risk of mortality until 70 days postoperatively (HR 1.14, 95% CI 1.01-1.28, p=0.029). For radiation interval, mortality escalated continuously with each additional day of delay, plateauing at 55 days (HR 1.25, 95% CI 1.11-1.41, p<0.001). Delays beyond these change points were not associated with further survival decrements.

Conclusion

Increasing delays in postoperative interval and radiation interval are independently associated with escalating mortality risk, which plateaus beyond certain thresholds. Delays in initiating therapy are conversely eclipsed in importance when appraised in conjunction with the entire treatment course. Such findings may redirect focus to streamlining intervals most sensitive to delays when considering survival burden.

INTRODUCTION

Advances in subspecialty head and neck cancer care have led to increased complexity in execution. Multidisciplinary evaluation, modern surgical reconstruction, and sophisticated radiation planning have individually improved outcomes^1-3, but at the potential cost of lengthy delays. Timeliness of care has become progressively difficult to reconcile,⁴ as delays in one phase may cascade into disruption of another. For example, comprehensive treatment planning could engender stage progression; extended surgical recovery could postpone adjuvant radiation; systemic therapy could trigger toxicities that precipitate treatment breaks. All such scenarios may unintentionally result in poor oncologic outcomes.

Prior studies have explored diagnosis to treatment interval (DTI)^5,6, postoperative interval^7,8, or radiation interval⁹, but not considered them collectively. Delays in each interval may individually worsen prognosis, but their comparable importance and magnitude of effect remain poorly defined. In addition, the impact of treatment interval prolongation likely has a complex, non-linear relationship with survival, which is rarely considered. We examine the relative impact of these interwoven treatment intervals on head and neck cancer patients completing definitive surgery with full-course radiation, using multivariable analyses modeling treatment intervals as continuous, non-linear functions.

METHODS

Data Source

Data were acquired from the National Cancer Database (NCDB), a tumor registry maintained by the American Cancer Society and the Commission on Cancer of the American College of Surgeons. The NCDB records data from more than 1500 hospitals, capturing roughly 70% of all cancer cases treated in the United States. All current NCDB head and neck cancer participant user files were evaluated, covering cases diagnosed between 2004 to 2013. This study was deemed exempt by the Cedars-Sinai institutional review board.

Patient Selection

Adult patients with biopsy-proven head and neck squamous cell carcinoma (HNSCC) of the oral cavity, oropharynx, larynx, and hypopharynx (ICD-03 8050-8054) were identified (Figure S1). To avoid confounding effects of subtherapeutic regimens, we focused strictly on patients who completed curative-intent surgery and full-course adjuvant radiation. Patients with incomplete staging, treatment, or follow-up data were excluded. Patients with clinical or pathologic distant metastasis were eliminated.

Patients with definitive surgery >365 days after diagnosis, or who did not have definitive surgery, were excluded. To eliminate patients who did not complete full-course radiation, the following criteria were required: minimum dose of 59.4 Gy, minimum of 30 fractions administered, and minimum radiation interval of 40 days (equivalent of 6 weeks of radiation treatment). Patients who received systemic therapy >14 days before or >14 days after starting radiation were excluded, to eliminate atypical chemotherapy regimens. Cases where it was unknown if systemic therapy was given were excluded. Patients with unknown, zero, or negative time intervals (diagnosis to treatment interval, postoperative interval, radiation treatment interval) were excluded, and a minimum postoperative interval of 14 days was required. To avoid early recurrences due to delayed radiation, cases with >180 days postoperative interval were excluded. Diagnosis to treatment interval was defined as the time from diagnosis until surgery. Postoperative interval was defined as the window from surgery to the first day of radiation. Radiation interval was defined as the time from the first to the last day of radiation. Treatment package time was defined as the sum of postoperative interval and radiation interval.

Statistical Methods

Missing data mechanism for the variables with missing values - race, insurance, income, education, urban/rural, great circle distance, ENE and margins (missing rates: 35.1% for ENE; 13.8% for urban/rural; and 0.9% - 3.4% for others) was examined using the method proposed by Little.¹⁰ We found that the data were not missing completely at random. To reduce the chance of bias from missing data, missing values were imputed using fully conditional specification (FCS) implemented by the multivariate imputation by chained equations (MICE) algorithm under the missing at random (MAR) assumption.¹¹ We generated 30 complete data sets, which were analyzed separately and then results were combined using the formula given in Rubin.¹²

Univariate associations between variables were examined with Wilcoxon rank-sum test for continuous variables and chi-square test for categorical variables. The primary outcome was overall survival (OS) calculated from diagnosis to the date of death or censored at last follow-up. Median follow-up was calculated using the reverse Kaplan-Meier method.¹³ Survival functions were estimated by the Kaplan-Meier method and compared using a log-rank test.¹⁴ Univariate and multivariable survival analyses were carried out using a Cox proportional hazards model.¹⁵ Multivariable analyses were performed using a stepwise variable selection procedure based on Akaike Information Criterion (AIC).¹⁶ Multivariable models with the lowest AIC values were selected as final models. The proportional hazards assumption was assessed graphically and analytically with scaled Schoenfeld residuals.¹⁷ The set S of covariates included in the multivariable model were common to all models fitted to the 30 imputed data sets. Each multivariable model from the 30 imputed data sets had between 0 and 1 additional covariates to the list in the set S. Likelihood ratio tests were carried out to compare each full model to the reduced model that has the set S of covariates and the results were not statistically significant.

Postoperative interval, radiation interval, diagnosis to treatment interval, radiation dose, and age were modeled using restricted cubic spline functions allowing for their non-linear associations with OS. The optimal number of knots was chosen based on the lowest AIC, and knots were placed at fixed percentiles of each variable (i.e., 10^th, 50^th, and 90^th percentiles for three knots; 5^th, 35^th, 65^th, and 95^th percentiles for four knots; and 5^th, 27.5^th, 50^th, 72.5^th, and 95^th percentiles for five knots).¹⁸ As a result, three knots were placed at 60, 63, and 70 for radiation dose; four knots at 28, 43, 55, and 93 for postoperative interval, at 41, 45, 50, and 65 for radiation interval, and at 7, 23, 37, and 76 for diagnosis to treatment interval; and five knots at 42, 53, 59, 66, and 78 for age. Estimated associations were illustrated with smoothed restricted cubic spline plots of the natural logarithm of adjusted hazard ratio versus postoperative interval and radiation interval, with 14 and 40 as the reference levels, respectively. Hazard ratio was estimated with a Cox proportional hazards model, which is adjusted for age with 5 knots, gender, region, insurance status, income, Charlson/Deyo comorbidity index, transfer to academic, tumor site, AJCC T-Classification, N-classification, extranodal extension, margins, postoperative chemotherapy, and radiation dose with 3 knots. Change points in postoperative interval and radiation interval were further estimated with piecewise linear regression models.¹⁹

Statistical analyses were performed using SAS 9.4 (SAS Institute, Inc., Cary, North Carolina) and R package version 3.4.1 (mice, rms, survival, segmented libraries) with two-sided tests and a significance level of 0.05.

RESULTS

Patient cohort

A total of 15,064 patients met inclusion criteria (Figure S1). Median follow-up time was 54.3 months (95% CI 53.6-55.2). Mean diagnosis to treatment interval (DTI) was 34.5 days (SD ± 24.2), mean postoperative interval was 52.9 days (SD ± 21.0), and mean radiation interval was 49.8 days (SD ± 12.2) (Figure 1, Table S1).

Figure 1.

Treatment intervals in study patient cohort.

Treatment intervals

In univariate analysis, all three treatment intervals were significantly associated with mortality (p<0.001) (Table 1). After adjustment for co-variates, DTI fell out of the model and held no prognostic value (Figure 2a-b, Figure S2), while prolonged postoperative interval (p<0.001) and radiation interval (p<0.001) remained strongly associated with worse overall survival (OS) (Figure 2c-f).

Table 1.

Univariate and multivariable analyses of overall survival.

Variable	Univariate			Multivariable
	N	HR (95% CI)	P-value	HR (95% CI)	P-value
Postoperative interval (Days)a	15064	-	<.001	-	<.001
Radiation interval (Days)a	15064	-	<.001	-	<.001
Diagnosis to tx interval (Days)a	15064	-	<.001	†
Age (Years) a	15064	-	<.001	-	<.001
Gender
Male	11040	1 (Reference)		1 (Reference)
Female	4024	1.06 (1.00-1.12)	0.043	0.90 (0.85-0.95)	<.001
Race b			<.001*
White	13221	1 (Reference)
Black	1399	1.28 (1.18-1.38)	<.001	†
Other	444	0.96 (0.81-1.13)	0.597
Facility type
Non-Academic	6767	1 (Reference)
Academic	8297	1.00 (0.95-1.05)	0.974	†
Facility volume
Low volume ≤ 75th percentile	11272	1 (Reference)
High volume > 75th percentile	3792	1.05 (0.99-1.11)	0.134	†
Region			<.001*		0.077*
East	3431	1 (Reference)		1 (Reference)
South	4682	1.21 (1.13-1.30)	<.001	1.10 (1.02-1.18)	0.016
Midwest	4817	1.07 (1.00-1.15)	0.060	1.04 (0.97-1.12)	0.291
West	2134	1.00 (0.91-1.09)	0.923	1.02 (0.93-1.12)	0.681
Insurance status b			<.001*		<.001*
Private	7257	1 (Reference)		1 (Reference)
Uninsured	868	1.60 (1.43-1.79)	<.001	1.22 (1.08-1.37)	0.001
Medicaid	1777	1.78 (1.64-1.93)	<.001	1.32 (1.21-1.44)	<.001
Medicare	4871	2.08 (1.97-2.21)	<.001	1.32 (1.22-1.43)	<.001
Other government	291	1.56 (1.30-1.88)	<.001	1.21 (1.00-1.46)	0.046
Zip-code level median household
Income b			<.001*		0.028*
< $38K	2772	1 (Reference)		1 (Reference)
$38K-$63K	7842	0.83 (0.78-0.89)	<.001	0.94 (0.88-1.00)	0.062
> $63K	4450	0.67 (0.62-0.72)	<.001	0.90 (0.83-0.97)	0.008
Zip-code level Education b			<.001*
Low: >21% no HS graduation	2568	1 (Reference)
Med: 7%-21%	9047	0.91 (0.85-0.97)	0.005	†
High: ≤7% no HS graduation	3449	0.75 (0.69-0.81)	<.001
Urban/Rural b			0.001*
Metro: ≥250K	12321	1 (Reference)
Urban: 2.5K-20K	2444	1.13 (1.06-1.22)	<.001	†
Rural: ≤2.5K	299	1.12 (0.94-1.35)	0.204
Great circle distance b			0.069*
<10mi	6206	1 (Reference)
10-20mi	3190	0.91 (0.85-0.97)	0.005	†
21-50mi	3197	0.97 (0.90-1.03)	0.325
51-100mi	1488	1.01 (0.92-1.11)	0.851
>100mi	983	0.98 (0.88-1.10)	0.726
Charlson/Deyo comorbidity index			<.001*		<.001*
0	11662	1 (Reference)		1 (Reference)
1	2718	1.37 (1.29-1.46)	<.001	1.14 (1.07-1.22)	<.001
≥ 2	684	1.97 (1.78-2.19)	<.001	1.52 (1.37-1.69)	<.001
Transfer to academic
No	9933	1 (Reference)		1 (Reference)
Yes	5131	0.94 (0.89-1.00)	0.035	0.92 (0.87-0.97)	0.003
Year of diagnosis			0.071*
2004-2006	3139	1 (Reference)
2007-2009	3852	0.95 (0.88-1.01)	0.115	†
2010-2013	8073	0.93 (0.87-0.99)	0.023
Tumor site			<.001*		<.001*
Oral cavity	7176	1 (Reference)		1 (Reference)
Oropharynx	4801	0.36 (0.33-0.39)	<.001	0.38 (0.35-0.42)	<.001
Larynx	506	1.22 (1.08-1.37)	0.001	0.82 (0.72-0.93)	0.002
Hypopharynx	2581	0.96 (0.90-1.03)	0.252	0.76 (0.71-0.82)	<.001
T-Stage			<.001*		<.001*
T1	3223	1 (Reference)		1 (Reference)
T2	4603	1.71 (1.56-1.87)	<.001	1.46 (1.34-1.60)	<.001
T3	2277	2.70 (2.45-2.97)	<.001	2.00 (1.81-2.21)	<.001
T4	4961	3.01 (2.77-3.28)	<.001	2.01 (1.83-2.20)	<.001
N-Stage			<.001*		<.001*
N0	4492	1 (Reference)		1 (Reference)
N1	2749	1.01 (0.93-1.09)	0.796	1.30 (1.19-1.41)	<.001
N2	7548	1.26 (1.19-1.34)	<.001	1.77 (1.62-1.94)	<.001
N3	275	1.16 (0.96-1.42)	0.128	2.35 (1.88-2.93)	<.001
Extranodal extension (ENE) b
ENE(−)	10363	1 (Reference)		1 (Reference)
ENE(+)	4701	1.43 (1.28-1.61)	<.001	1.34 (1.15-1.56)	<.001
Margins b
Negative	11728	1 (Reference)		1 (Reference)
Positive	3336	1.17 (1.10-1.25)	<.001	1.28 (1.20-1.36)	<.001
Postoperative chemotherapy
No	7482	1 (Reference)		1 (Reference)
Yes	7582	0.94 (0.89-0.99)	0.015	0.84 (0.79-0.90)	<.001
Radiation dose (Gy) a	15064	-	<.001	-	<.001

15064 observations were used in the multivariable model.

^aModeled using restricted cubic spline functions with three knots at 60, 63, and 70 for radiation dose; four knots at 28, 43, 55, and 93 for postoperative interval, at 41, 45, 50, and 65 for radiation interval, at 7, 23, 37, and 76 for diagnosis to treatment interval; and five knots at 42, 53, 59, 66, and 78 for age.

^bMissing data were imputed by multiple imputation.

^*Overall p-value for categorical variables with more than two levels.

^†Variables dropped out of the model.

Figure 2.

(A) Unadjusted and (B) adjusted Kaplan-Meier estimates of overall survival, stratified by diagnosis to treatment interval. (C) Unadjusted and (D) adjusted Kaplan-Meier estimates of overall survival, stratified by postoperative interval. (E) Unadjusted and (F) adjusted Kaplan-Meier estimates of overall survival, stratified by radiation interval.

Given their nonlinear relationship, restricted cubic spline functions were utilized to model the relationship between treatment interval and survival (Table 2). For postoperative interval, a reference level was set at 14 days (defined as the minimum recovery period before starting adjuvant radiation). No significant detriment was seen up to 40 days after surgery. However, mortality risk beyond this time point escalated with each day of delay (HR 1.004, 95% CI 1.000-1.008, p=0.029), plateauing at 70 days (HR 1.14 for 30 days increment, 95% CI 1.01-1.28, p=0.029) (Figure 3a). Further postoperative delays did not appear to worsen the risk of death. The estimated 5-year OS for the <40d, 40-70d, and >70d subgroups was 66.5%, 56.8%, and 50.0%, respectively.

Table 2.

Summary of hazard ratios for postoperative interval and radiation treatment interval, stratified by change point

Variable	Univariate		Multivariable a
	HR (95% CI)	P-value	HR (95% CI)	P-value
Postoperative interval
≥ 71 days	1.001 (0.998-1.004)	0.575	1.001 (0.998-1.004)	0.574
40 - 70 days	1.012 (1.008-1.016)	<.001	1.004 (1.000-1.008)	0.029
< 40 days	1.021 (1.010-1.032)	<.001	1.006 (0.994-1.017)	0.301
Radiation interval
≥ 55 days	1.001 (0.999-1.003)	0.272	1.000 (0.998-1.002)	0.771
< 55 days	1.027 (1.019-1.035)	<.001	1.016 (1.007-1.025)	<.001

Hazard ratio (HR) is expressed as 1-unit increment; Missing data were imputed by multiple imputation.

^aVariables included in multivariable models were age with 5 knots at 42, 53, 59, 66 and 78, gender, region, insurance status, income, Charlson/Deyo comorbidity index, transfer to academic, tumor site, AJCC T-Classification, N-classification, extranodal extension, margins, postoperative chemotherapy, radiation dose with 3 knots at 60, 63 and 70, and radiation interval with 4 knots at 41, 45, 50 and 65 or postoperative interval with 4 knots at 28, 43, 55 and 93.

Figure 3.

Natural logarithm of adjusted hazard ratio (HR) with increasing postoperative interval and radiation treatment interval. (A) Black solid line represents smoothed restricted cubic spline plot of the natural logarithm of predicted adjusted HR versus postoperative interval (days), with reference value of 14. Blue vertical lines represent the estimated change points at 40 and 70 days. (B) Black solid line represents smoothed restricted cubic spline plot of the natural logarithm of predicted adjusted HR versus radiation treatment interval (days), with reference value of 40. Blue vertical line represents the estimated change point at 55 days. Gray dashed lines represent estimated 95% CIs of the predicted HRs.

For radiation interval, a reference level was set at 40 days (equivalent to a therapeutic 6-week adjuvant radiation course). Risk of death increased continuously with each day of delay (HR 1.016, 95% CI 1.007-1.025, p<0.001), up to a change point of 55 days (HR 1.25 for 14 days increment, 95% CI 1.11-1.41, p<0.001) (Figure 3b). Further prolongation of the radiation interval had no additional impact on survival. The estimated 5-year OS for the <55d and ≥55d subgroups was 59.9% and 50.8%, respectively.

Although postoperative and radiation intervals were evaluated as independent variables, they are often considered compositely as treatment package time. We further investigated the window of negative impact in this compound interval, if any, with a reference level set at 54 days (minimum 14 days postoperative interval with minimum 40 days radiation interval). No significant mortality risk was noted if treatment was completed within 84 days of surgery (Figure S3). However, mortality risk increased with each day of delay beyond 84 days (HR 1.007, 95% CI 1.004-1.010, p<0.001), plateauing at 122 days (HR 1.28 for 38 days increment, 95% CI 1.15-1.44, p<0.001). Further delays did not appear to worsen the risk of death. The <84d, 84-122d, and >123d subgroups comprised 16.7%, 68.1%, and 15.2% of patients, respectively; the estimated 5-year OS was 70.0%, 57.7%, and 47.9%, respectively (Figure S4).

Given that prolonged DTI was not associated with survival, we explored additional factors influenced by treatment initiation delays that might secondarily affect mortality. Subgroup analysis was performed with early-stage patients (T1-2N0 cases), to assess if stage progression due to such delays conferred undue risk. In this cohort, DTI remained nonsignificant on multivariable analysis, while postoperative interval and radiation interval remained significant (Table S2). Similar results were observed with subgroup analysis of HPV(+) oropharyngeal cancer patients (Table S3).

DISCUSSION

In this analysis, we examined the collective impact of treatment intervals spanning diagnosis to therapy completion on head and neck cancer patients undergoing surgery and adjuvant radiation. Using a granular multivariable approach, we demonstrated progressively worsening outcomes with escalating delays in both postoperative interval and radiation interval, but only during specific windows of time. For postoperative interval, mortality risk began to increase beyond 5.7 weeks after surgery, plateauing at 10 weeks; for radiation interval, mortality risk began increasing immediately with any delay, plateauing at 7.9 weeks from starting radiation. The cumulative mortality risk noted at these thresholds render such delays as important surrogates for poor prognosis: their values are roughly comparable to the hazard conferred by classic adverse factors such as positive margins (HR 1.28, 95% CI 1.20-1.36) or extranodal extension (HR 1.34, 95% CI 1.15-1.56).

Unlike these inflexible tumor-intrinsic elements, it is compelling that treatment-interval factors are potentially modifiable. Disturbingly, only 16.7% of patients in this study had a composite treatment package time of 83 days or less, the minimum threshold before mortality risk began to rise; even less were able to complete therapy on time (by 54 days). Minimizing treatment delay thus remains a formidable yet targetable area for improving head and neck cancer outcomes. Notably, we restricted our analysis only to ideal cases where curative-intent surgery and full-course radiation were completed: patients with subtherapeutic, incomplete, or truncated treatment regimens may be expected to do far worse.

Execution of multimodality head and neck cancer care is often undermined by the practical realities of treatment. Factors such as travel logistics, postsurgical recovery, treatment complications, lack of supportive ancillary services, and poor performance status may produce delays in overall treatment time. These delays in turn influence survival via established biological mechanisms involving accelerated repopulation of remnant cancer cells, diminishing radiation efficacy and breeding tumor radioresistance.^20,21 The National Comprehensive Cancer Network (NCCN) Guidelines²² have recommended that the postoperative interval be <6 weeks, based on a meta-analysis of studies that arbitrarily selected this threshold based on poorer locoregional control.²³ Others have emphasized treatment package time of approximately 100 days.^7,24 While our results support these cutpoints, they also quantify a continuum of risk via an agnostic cut-point evaluation, with no detriment to survival observed if radiation was initiated within 40 days of surgery. This naturally coincides with a reasonable window for healing after surgery. On the other hand, any prolongation of the radiation interval appears to adversely affect survival, worsening well beyond binary thresholds currently defined in guidelines. Finally, incremental mortality risk plateaus with each interval in our analysis, suggesting that colossal delays are less consequential: the damage is already done.

Although prolonged diagnosis to treatment interval (DTI) was associated with worse survival when considered in isolation, its prognostic value disappeared in multivariable analysis when examined in combination with the impact of postoperative interval and radiation interval. Our analysis helps clarify the conflicting body of literature surrounding the relevance of DTI. Prior studies have failed to consistently identify an effect from DTI delays, including in head and neck, pancreatic, lung, breast, colorectal, and pancreatic malignancies.^25-28 While a meaningful percentage of HNSCC cases exhibit volumetric progression with delay prior to surgery^29,30, it is unclear if this translates to worsened outcomes.³¹ One analysis of HNSCC patients treated between 2003-2005 detected a relationship with delayed DTI and decreased survival only beyond 60 days,⁵ while another Taiwanese study found DTI beyond 20 days increased mortality risk.⁶ However, these cohorts comprised a heterogeneous mix of patients treated nonsurgically and surgically, with and without radiation, and without certainty if therapy was completed. In addition, no adjustment was made for other treatment intervals, which we demonstrate as pivotal (Figure 2, Figure S2). These differences in methodology likely account for the contrasting results observed in our study. Moreover, early-stage patients (11.5% of our cohort) did not perform disproportionately worse with DTI delays, suggesting that presumed stage progression from delays does not impair survival at a population level. While a massive DTI delay should practically confer risk, we excluded patients with a DTI of greater than one year to avoid the influence of extreme outlier cases on our analysis. Our results appear to offer flexibility to complete time-consuming but valuable treatment planning tasks, such as multidisciplinary tumor board discussion, ancillary service evaluations, second opinions at other institutions, and transfers to tertiary specialized centers.

Several caveats deserve mention, including the study’s observational design and the lack of information on disease-specific survival. The results may not fully translate to definitive chemoradiation patients, or to surgical patients who did not receive radiation. It is conceivable that patients with poor performance status or aggressive tumors (“bad actors”) may be more prone to treatment delays and drive poorer outcomes: this remains difficult to correct for, outside of our adjustment for comorbidity index and TNM stage. Likewise, particularly aggressive patients fell outside our strict criteria if they could not finish treatment; while their subtherapeutic regimens would confound the analysis (by having paradoxically shortened treatment intervals), the significance of our findings for this subgroup remains unknown. Finally, certain factors correlating with outcome, including smoking status, chemotherapy type, perineural invasion, and recurrence status were not available and may have some degree of unmeasured impact. Nonetheless, we believe that our results are the strongest and most relevant to date for characterizing the relative impact of treatment delays on head and neck cancer patients.

In summary, protracted postoperative and radiation intervals significantly impair survival in HNSCC patients completing surgery and full-course radiation, with escalating mortality bounded inside specific windows of time. Despite the harm imparted by treatment delays, only a minority of patients completed treatment within the optimal treatment package time window. Delays in treatment initiation conversely fail to influence survival when considered in concert with other phases. Taken in perspective, the mortality risk conferred by delayed treatment package time is similar in magnitude to the benefit from therapeutic factors such as concomitant chemotherapy, which enjoys wide acceptance for high-risk patients. The meaningful hazard from treatment delays therefore support the critical importance of efficient interdisciplinary coordination as itself a therapeutic modality, to anticipate setbacks and address them as aggressively as the cancer itself. Such a focus may yield an underestimated opportunity to streamline management and advance patient outcomes.