Optimal radiotherapy dose in cervical esophageal squamous cell carcinoma patients treated with definitive concurrent chemoradiotherapy: A population based study.

BACKGROUND: The optimal radiotherapy dose for locally advanced cervical esophageal squamous cell carcinoma (C-ESqCC) treated with definitive concurrent chemoradiotherapy (dCCRT) is unclear. Here, we aimed to compare the survival of those treated with high dose versus standard dose via a population based approach.
METHODS: Eligible C-ESqCC patients diagnosed between 2011 and 2017 were identified via the Taiwan Cancer Registry. We used propensity score (PS) weighting to balance observable potential confounders between groups. The hazard ratio (HR) of death and incidence of esophageal cancer mortality (IECM) were compared between high (60-70 Gy) and standard dose (50-50.4 Gy). We also evaluated the outcome in supplementary analyses via alternative approaches.
RESULTS: Our primary analysis consisted of 141 patients in whom covariates were well balanced after PS weighting. The HR of death when high dose was compared with standard dose was 0.65 (95% confidence interval [CI]: 0.4-1.03, p = 0.07). The HR of IECM was 0.74 (p = 0.45). The HR of OS remained similarly insignificant in supplementary analyses.
CONCLUSIONS: We observed a trend in favor of high radiotherapy dose versus standard dose for C-ESqCC treated with dCCRT in this population-based nonrandomized study. Further studies are needed to confirm the findings of the study.

INTRODUCTION

Esophageal cancer is one of the leading most common cause of cancer mortality worldwide including Taiwan. The most common histology is adenocarcinoma in western countries whereas it is squamous cell carcinoma (SqCC) in Asia. ,

For cervical ESqCC (C‐ESqCC), definitive concurrent chemoradiotherapy (dCCRT) is commonly suggested in treatment guidelines, especially in patients with locally advanced disease. , , , However, the optimal radiotherapy dose is unknown. For the commonly‐seen thoracic EqCC, the standard dose is 50–50.4 Gy at standard fractionation, as suggested by the North American guidelines. This dose has been established by randomized controlled trials (RCTs) , although it is still debated in the literature , and in some guidelines. , For the relatively rare C‐ESqCC, “higher doses may be appropriate for tumors of the cervical esophagus” is stated in the North American guidelines. The rationale might be that the treatment for C‐ESqCC has been adopted from regimens for esophageal SqCC, as well as head‐and‐neck SqCC, and therefore 60–70 Gy has been the previously suggested dose. , Although C‐ESqCC was eligible in previous radiotherapy doses in RCTs, , it accounted for less than 10% of participants and results specific to C‐ESqCC have not been reported. , However, high dose dCCRT has been previously advocated in the literature.

Two comparable effectiveness studies were reported in the above mentioned review paper published in 2020 regarding cervical esophageal cancer patients treated with different radiotherapy dose groups and all reported better outcomes in the higher dose group. , However, both were single institutional studies with sample size less than one hundred. Other studies have also reported higher radiotherapy dose as a better prognostic factor.

Due to the above limitations and uncertainty regarding the optimal radiotherapy dose, our study aimed to compare the survival of high versus standard radiotherapy dose for C‐ESqCC patients treated with dCCRT via a population‐based approach.

METHODS

Data

We used the Taiwan Cancer Registry (TCR) as the data source in this study. The quality of the TCR is reported to be one of the highest‐quality cancer registries in the world. , Our study was approved by the review committee (National Health Research Institute EC1090502‐E).

Study population and intervention

We identified C‐ESqCC patients diagnosed within 2011–2017 from TCR via the International Classification of Disease for Oncology third edition (ICD‐O‐3) site and histology codes. We defined locally advanced disease as patients with clinical stage cT2‐4N0M0 or cT1‐4N1‐3M0 by the seventh American Joint Committee on Cancer (AJCC). We only included adult patients age 18–70. We excluded those with multiple treatment records to ensure data quality. We also excluded those with prior cancer(s).

Regarding intervention, we further identified patients treated with dCCRT using external beam intensity‐modulated radiotherapy in conventional dose fractionation (1.8–2 Gy/fraction) according to records in TCR then selected those treated with high radiotherapy dose (group A, 60–70 Gy) or standard dose (group B, 50–50.4 Gy). We allowed ± 5% tolerance in RT dose for both groups. We defined dCCRT as patients treated with concurrent systemic therapy and radiotherapy without surgery. These inclusion/exclusion criteria were modified from our clinical and research experiences as well as previous studies. ,

Covariates

We included the following covariates as modified from recent relevant studies and our clinical and research experiences. , , , , Patient demographics (age, gender, residency), patient characteristics (body mass index [BMI], drinking, smoking), disease characteristics (T and N and overall stage), treatment characteristics (peri‐CCRT systemic therapy, use of image‐guided radiotherapy [IGRT], radiotherapy break) and the use of positron emission tomography (PET) were defined as follows. Patient residency region was classified as “northern Taiwan” or “non‐north”. , Smoking, drinking and the use of PET or IGRT were classified as yes or no. The details of IGRT (such as 2D or 3D) was not available in TCR. The clinical T stage was classified as “1–2” versus “3–4”. The clinical N stage was classified as “0–1” or “2–3”. Overall staging was grouped as 2 versus 3. Those patients who received pre‐ or post‐CCRT systemic therapy was classified as “yes” for “peri‐CCRT systemic therapy” (induction or adjuvant or both) whereas those without were classified as “no”. Those patients with radiotherapy prolongation of more than one week were classified as “yes” for “radiotherapy break,” whereas those without were classified as “no”.

Statistical and supplementary analyses

The primary outcome of interest was overall survival (OS). We also evaluated the impact of intervention (high vs. standard dose) on incidence of esophageal cancer mortality (IECM). We adopted a propensity score (PS) approach and used PS weighting (PSW) as the framework for analyses as advocated in the literature. , , , We estimated the probability of receiving a high radiotherapy dose (vs. standard dose) with a logistic regression model based on all the above covariates, and then assessed the balance of covariates between groups after PSW using overlap weight via the standardized difference (SDif) rather than the chi‐square or t‐test. , We compared the hazard ratio (HR) of death between the group A and group B groups during the entire follow‐up period via Cox proportional hazards model in the weighted sample for point estimation and used the bootstrap method to estimate the 95% confidence interval (95% CI). , , We used E‐value to assess the robustness of our finding regarding potential unmeasured confounder(s) as suggested in the literature because a PS approach is only be valid under the assumption of no unmeasured confounder(s). We took a competing risk approach to compare IECM between groups. We performed two supplementary analyses (SA) using alternative analytic approaches. We did SA‐1 via using PS matching (PSM) by constructing a 1:1 matched cohort and performed the analyses as described in the literatures. , We did SA‐2 by using Cox regression method as suggested by the reviewer during revision. We used software SAS 9.4 (SAS Institute) and R version 3.6.2 (R Development Core Team) for statistical analyses.

RESULTS

Study population

We identified 141 patients (114 for group A and 27 for group B) as our primary study population as shown in Figure 1. We achieved covariate balance after PSW although some imbalance was seen before PSW as shown in Table 1. After a median follow‐up of 19 months (range 3–102) with 21 months (range 3–102) for group A, as well as 13 months (range 3–50) for group B, death was observed in 82 patients in group A and in 23 patients in group B. The median follow‐up was 61 months (range 29–102) for survivors.

FIGURE 1

STROBE study flowchart and the number of individuals at each stage of the study. 1. We only included those treated (class 1–2) with only one record to ensure data consistency. 2. The Seventh American Joint Committee on Cancer staging clinical stage T2‐4N0M0 or T1‐4N1‐3M0. 3. With concurrent systemic therapy and radiotherapy without surgery. 4. Conventional fractionated external beam radiotherapy dose (high, 60–70 Gy vs. standard, 50–50.4 Gy) at 1.8–2 Gy/fraction, ± 5% tolerance in dose. 5. Without missing information in the TCR and death registry

TABLE 1

Patient characteristics of the study population in the primary analysis

	Group A (high dose) (n = 114)		Group B (standard dose) (n = 27)		Standardized difference (rounded) a
	Number or mean (SD) a	(%)a	Number or mean (SD) a	(%)a	Before PSW	After PSW
Age (year)	53.98 (7.29)		54.52 (6.66)		0.077	≈0
Gender
Female	13	(11)	1	(4)	0.295	≈0
Male	101	(89)	26	(96)
Residency
Non‐north	72	(63)	16	(49)	0.080	≈0
North	42	(37)	11	(41)
BMI	21.25 (3.73)		21.43 (4.02)		0.047	≈0
Drinking
No	16	(14)	3	(11)	0.088	≈0
Yes	98	(86)	24	(89)
Smoking
No	15	(13)	4	(15)	0.048	≈0
Yes	99	(87)	23	(85)
T stage
1–2	13	(11)	4	(15)	0.101	≈0
3–4	101	(89)	23	(85)
N stage
0–1	60	(53)	13	(48)	0.090	≈0
2–3	54	(47)	14	(52)
Overall stage
2	18	(16)	5	(19)	0.072	≈0
3	96	(84)	22	(81)
Peri‐CCRT systemic therapy
No	73	(64)	19	(70)	0.135	≈0
Yes	41	(36)	8	(30)
Use of IGRT
No	84	(74)	14	(52)	0.464	≈0
Yes	30	(26)	13	(48)
Radiotherapy break
No	91	(80)	22	(81)	0.042	≈0
Yes	23	(20)	5	(19)
Use of PET
No	30	(26)	13	(48)	0.464	≈0
Yes	84	(74)	14	(52)

Abbreviations: BMI, body mass index; CCRT, concurrent chemoradiotherapy; IGRT, image‐guided radiotherapy; PET, positron emission tomography; PSW, propensity‐score weighting; SD, standard deviation.

Rounded.

Primary analysis

The overlap weights adjusted OS curve are shown in Figure 2. The one/two/three/four‐year OS rates for both groups were 59%, 48%, 38% and 32% (group A) and 59%, 42%, 23% and 7% (group B), respectively. When group A (high dose) was compared to group B (standard dose), the HR of death was 0.65 (95% confidence interval [95% CI]: 0.4–1.03, p = 0.07). The observed HR of 0.65 for OS could be explained by an unmeasured confounder that was associated with both selections of treatment (group A vs. B) and outcome (live vs. death) by a risk ratio of 2.03 (E‐factor) fold each, but weaker confounding could not. The HR of IECM was 0.74 (95% CI: 0.33–1.64, p = 0.45).

FIGURE 2

The overlap weight‐adjusted overall survival curve (in years) in the primary analysis

Supplementary analyses

In SA‐1 (PSM), imbalance (SDif > 0.25) was observed for three covariates (gender, use of IGRT, use of PET) before PSM, whereas only one covariate (use of PET) remained imbalanced after PSM, as shown in Tables 1 and 2. When group A (high dose) was compared to group B (standard dose), the HR of death was 0.63 (95% CI: 0.34–1.16, p = 0.13). The OS curve for the PSM cohort is shown in Figure 3. The HR of IECM was 0.66 (95% CI: 0.35–1.23, p = 0.19).

TABLE 2

Patient characteristics of the study population in the supplementary analysis

	Group A (high dose) (n = 27)		Group B (standard dose) (n = 27)
	Number or mean (SD) a	(%)a	Number or mean (SD) a	(%)a	Standardized difference (rounded) a
Age (year)	54.41 (5.54)		54.52 (6.66)		0.018
Gender
Female	1	(4)	1	(4)	0
Male	26	(96)	26	(96)
Residency
Non‐north	18	(67)	16	(59)	0.154
North	9	(33)	11	(41)
BMI	21.06 (3.39)		21.43 (4.02)		0.100
Drinking
No	3	(11)	3	(11)	0
Yes	24	(89)	24	(89)
Smoking
No	3	(11)	4	(15)	0.110
Yes	24	(89)	23	(85)
T stage
1–2	3	(11)	4	(15)	0.110
3–4	24	(89)	23	(85)
N stage
0–1	12	(44)	13	(48)	0.074
2–3	15	(56)	14	(52)
Overall stage
2	4	(15)	5	(19)	0.100
3	23	(85)	22	(81)
Peri‐CCRT systemic therapy
No	21	(78)	19	(70)	0.170
Yes	6	(22)	8	(30)
Use of IGRT
No	17	(63)	14	(52)	0.226
Yes	10	(37)	13	(48)
Radiotherapy break
No	23	(85)	22	(81)	0.100
Yes	4	(15)	5	(19)
Use of PET
No	9	(33)	13	(48)	0.305
Yes	18	(67)	14	(52)

Abbreviations: BMI, body mass index; CCRT, concurrent chemoradiotherapy; IGRT, image‐guided radiotherapy; PET, positron emission tomography; PSW, propensity‐score weighting; SD, standard deviation.

Rounded.

FIGURE 3

The overall survival curve (in years) in the supplementary analysis (SA‐1)

In SA‐2 (Cox regression), we found group A (high dose) was associated with insignificantly lower HR of death (0.72, 95% CI: 0.43–1.21, p = 0.21) when compared with group B (standard dose) (Table 3). Because three covariates (gender, use of IGRT, use of PET) were imbalanced between groups before PS analyses, we also performed univariate and multivariate analyses as suggested by the reviewer during revision and found the results remained insignificant in multivariate analyses (HR of death 0.66, 95% CI: 0.41–1.07, p = 0.09) (Tables S1 and S2).

TABLE 3

Hazard ratio of death in SA‐2 (Cox regression)

	HR a	95% CI a	p‐value a
Dose	0.72	0.43–1.21	0.21
Age (year)	1.01	0.99–1.04	0.33
Gender	0.87	0.38–2.00	0.75
Residency	0.91	0.60–1.37	0.64
BMI	0.87	0.82–0.93	<0.0001
Drinking	1.67	0.75–3.68	0.21
Smoking	1.18	0.5–2.79	0.71
T stage	1.87	0.79–4.44	0.16
N stage	1.97	1.20–3.24	0.01
Overall stage	0.79	0.35–1.77	0.56
Peri‐CCRT systemic therapy	0.75	0.49–1.15	0.19
Use of IGRT	0.74	0.46–1.17	0.19
Radiotherapy break	1.83	1.07–3.15	0.03
Use of PET	0.68	0.44–1.06	0.09

Abbreviations: 95% CI, 95% confidence interval; BMI, body mass index; CCRT, concurrent chemoradiotherapy; HR, hazard ratio; IGRT, image‐guided radiotherapy; PET, positron emission tomography; SA, supplementary analysis.

Rounded.

DISCUSSION

In this population‐based nonrandomized study on C‐ESqCC patients treated with definitive concurrent chemoradiotherapy, there was a trend in favor of high radiotherapy dose versus standard dose with borderline significance (p = 0.06) in the primary analyses (PSW), although significance (p = 0.03) was observed in the supplementary analysis (PSM). To the best of our knowledge, this is the first population based study in C‐ESqCC.

Single arm studies , have previously investigated the effect of radiotherapy in cervical esophageal cancer. Furthermore, as mentioned previously, some single institutional studies have also reported the radiotherapy dose effect on these patients. , McDowell et al. reported that a high dose (analyzed as a continuous variable) was associated with a borderline improved OS (HR 0.97, p = 0.075). Kim et al. also reported that a high dose (≥59.4 Gy) was associated with reduced risk of death with HR 0.88 when compared to low dose (<59.4 Gy). We further searched PubMed in January 2021 using the keywords “(cervical esophageal cancer) AND (squamous cell carcinoma) AND ((radiation therapy dose) OR (radiotherapy dose))” and found no additional relevant studies. The findings in our study (a trend of reduced risk of death for high dose) was in line with these studies. ,

The interpretation of our study appears straightforward due to the general belief that C‐ESqCC should be treated more like other head‐and‐neck SqCC. However, it should be noted that the survival difference within median follow‐up (around one and a half year) was not obvious (see Figure 2 and subsection “Primary analysis” in results). Our study should also be interpreted with caution given its nonrandomized nature and borderline statistical significance. However, when we searched the trial registry [https://clinicaltrials.gov/] in March 2021, we did not find any relevant RCT specific for C‐ESqCC. Therefore, we believe that our study provides useful evidence regarding radiotherapy for C‐ESqCC while we await the results of more studies on this subject.

There were also limitations in our study. First,as in all nonrandomized studies, potential unmeasured confounder(s) such as patient social economic status, bodyweight loss or performance status, biomarkers, precision in staging work‐up, use of supportive care (such as nasogastric tube or enterostomy), chemotherapy regimen details, or radiotherapy tolerability or techniques (such as target delineation and elective nodal region) were not available due to data limitation, although we had used a PS approach to balance observed covariates and reported E‐value to assess the potential impact of potential unmeasured confounder(s). Although guidelines within individual institutions are encouraged by the government, there was a lack of general guidelines, so the treatment selection (high vs. standard radiotherapy) was at the discretion of the treating radiation oncologists (i.e., not randomly assigned). Second, the use of salvage therapy (especially surgery) may have impacted our primary endpoint (OS) but could not be evaluated due to data limitation in TCR. Finally, other endpoints, such as pattern of failure or quality of life in addition to OS used in our study, might also be relevant, but these were not investigated due to concern with regard to data availability.

In conclusion, we observed a trend in favor of high radiotherapy dose versus standard dose for C‐ESqCC treated with dCCRT in this population‐based nonrandomized study. Further studies (especially RCT) are needed to confirm our findings.

CONFLICT OF INTERESTS

All authors declare there are no conflicts of interest.

Table S1 univariate survival analyses of subgroups with large standardized difference in primary analyses before propensity score weighting

Table S2 multivariate survival analyses considering the covariates with large standardized difference in primary analyses before propensity score weighting

Click here for additional data file.