Impact of tumor volume enlargement after induction chemotherapy on subsequent radiotherapy in locally advanced nasopharyngeal carcinoma: A propensity‐score matching analysis

Abstract A small proportion of nasopharyngeal carcinoma (NPC) patients show resistance to induction chemotherapy (IC). This study sought to investigate the impact of tumor volume enlargement after IC on the dosimetric parameters of subsequent radiotherapy. The records of a total of 240 locally advanced NPC patients who received IC followed by concurrent chemoradiotherapy were retrospectively reviewed. Patients with a tumor volume enlargement of ≥10% and patients with a tumor volume reduction of ≥10% after induction chemotherapy were classified as the enlargement group and the control group, respectively. The dosimetric parameters of the planning target volumes (PTVs) and the organs at risk (OARs) were compared between the matched groups after propensity score matching (PSM). For the gross tumor volume of nasopharynx (GTVnx), 21 patients and 127 patients were classified as the enlargement group and the control group, respectively. After matching, 20 sub‐pairs of 40 patients were generated in the post‐PSM cohort. The GTVnx enlargement group exhibited no significant disadvantages in all of the dosimetric parameters, except in the planning organ‐at‐risk volume (PRV) of contralateral lens (Dmax, 722 cGy vs. 634 cGy, p = 0.041). For the gross tumor volume of lymph nodes (GTVnd), 44 patients and 144 patients were classified as the enlargement group and the control group, respectively. After matching, 39 sub‐pairs of 78 patients were generated in the post‐PSM cohort. The GTVnd enlargement group exhibited no significant disadvantages in all of the dosimetric parameters. Univariate and multivariate analyses showed that the enlargement of GTVnx and the enlargement of GTVnd were not independently associated with any of the dosimetric parameters. A tumor volume enlargement of ≥10% in GTVnx or GTVnd after induction chemotherapy has no significant impact on the dosimetric parameters of subsequent radiotherapy in locally advanced NPC.


Funding information
Univariate and multivariate analyses showed that the enlargement of GTVnx and the enlargement of GTVnd were not independently associated with any of the dosimetric parameters. A tumor volume enlargement of ≥10% in GTVnx or GTVnd after induction chemotherapy has no significant impact on the dosimetric parameters of subsequent radiotherapy in locally advanced NPC.

| INTRODUCTION
Nasopharyngeal carcinoma (NPC) is a common type of head and neck cancer, and radiotherapy is its principle treatment method due to the complicated anatomical location of the cancer and its high sensitivity to radiation. 1,2 Recently, induction chemotherapy (IC) followed by concurrent chemoradiotherapy has become a standard treatment strategy for locally advanced NPC according to the National Comprehensive Cancer Network (NCCN) guidelines (version 1.2020). Several well-designed randomized controlled studies have shown that IC can improve the survival outcomes of NPC patients. [3][4][5] However, it is noteworthy that although most NPC patients are sensitive to chemotherapy, a small proportion of patients show resistance to IC, which has a negative influence on patient survival. 6,7 In addition to the impact on patient survival, another potential consequence of tumor progression after IC is the influence on subsequent radiotherapy. The location of NPC is surrounded by many important structures, such as the brainstem, spinal cord, and optic nerves. 8,9 An increase in tumor volume after IC may have a significant influence on the dosimetric parameters of the subsequent radiotherapy plan. However, no publications have addressed this problem thus far. Therefore, we conducted this retrospective study to compare the radiotherapy plans of patients with tumor volume enlargement and patients with tumor volume reduction after IC using the propensity score matching (PSM) method.

| Patient selection
A total of 240 NPC patients were selected according to the following criteria: (a) locally advanced NPC (T1-2N1-3M0 or T3-4N0-3M0, according to the AJCC 8th staging system) with pathology confirmation; (b) treated with IC followed by concurrent chemoradiotherapy at our hospital from 2016 to 2019; (c) contrast-enhanced simulation CT and simulation MRI were performed before and after IC; and (4) the radiotherapy plan was available for review. Written informed consent was obtained from each subject.

| Induction chemotherapy
All of the patients received IC with docetaxel plus cisplatin (docetaxel 75 mg/m 2 day 1, cisplatin 75 mg/m 2 day 2) for two or three cycles. The chemotherapy cycle was repeated every 21 days. Adequate bone marrow function, liver function, and renal function were required before the start of each chemotherapy cycle. All of the patients underwent contrastenhanced simulation CT and simulation MRI at a 3-mm slice thickness with immobilization devices before IC. The gross tumor volume of nasopharynx (pre-IC GTVnx) and the gross tumor volume of lymph nodes (pre-IC GTVnd) were contoured with CT and MRI fusion images by a medical team consisting of radiation oncologists and radiologists.

| Concurrent chemoradiotherapy
Radiotherapy was delivered 3 weeks after the last cycle of IC, concurrent with cisplatin 80-100 mg/m 2 every three weeks or 30-40 mg/m 2 every week. 10 Contrast-enhanced simulation CT and simulation MRI at a 3-mm slice thickness with immobilization devices were performed again for the preparation of radiotherapy. The gross tumor volume of nasopharynx (post-IC GTVnx) and the gross tumor volume of lymph nodes (post-IC GTVnd) were contoured again by the same medical team.
The target volumes in the radiotherapy plan included the final GTVnx, the final GTVnd, the clinical target volume 1 (CTV1), and the clinical target volume 2 (CTV2) according to the recommendation of the international guideline for the delineation of the clinical target volumes for NPC. 11 The final GTVnx was defined as the summation of the pre-IC GTVnx and the post-IC GTVnx, which included all of the areas involved by the primary tumor before and after IC. The final GTVnd was defined as the post-IC GTVnd only. CTV1 and CTV2 were defined as the high-risk volume and the low-risk volume, respectively.
The radiotherapy planning techniques consisted of conventional intensity-modulated radiation therapy (IMRT) and tomotherapy. The conventional IMRT plans, which included the volumetric-modulated arc therapy and the step-and-shoot IMRT, were generated with the Eclipse treatment planning system (Eclipse version 11.3, Varian Medical Systems). The tomotherapy plans were generated with the TomoTherapy Planning Workstation (TomoHD version 2.0.7, Accuracy Inc.).

| Dosimetric comparisons
Patients with a tumor volume enlargement of ≥10% and patients with a tumor volume reduction of ≥10% after IC were classified as the enlargement group and the control group, respectively. PSM was adopted to control the balance between the enlargement group and its control group. Matching covariates in the score scale included T stage, N stage, plan type, pretreatment volume of GTVnx, and pretreatment volume of GTVnd. For the PTVs, the minimum coverage dose of 95% of the target (D95) was selected as the dosimetric parameter for comparisons in the post-PSM cohort. For the OARs, the maximum dose (Dmax) was adopted to evaluate the dosimetric differences of the brainstem, brainstem PRV, spinal cord, spinal cord PRV, optic chiasm, optic nerve, lens PRV, and pituitary between the matched groups in the post-PSM cohort. In addition, the relative volume receiving over 30 Gy (V30 Gy) and the relative volume receiving over 60 Gy (V60 Gy) were selected to evaluate the dosimetry of the parotid glands and temporal lobes, respectively.

| Statistical analyses
All of the statistical analyses were performed with SPSS (version 25, IBM SPSS Statistics). The comparisons of baseline characteristics between the enlargement group and the control group were made with the independent t test and chi-square test. Dosimetric comparisons between F I G U R E 1 A typical case of GTVnx enlargement (left) and its matched control case(right) in the post-PSM cohort. The red lines represent the contours of GTVnx before induction chemotherapy. The purple lines represent the contours of GTVnx after induction chemotherapy. (GTVnx = the gross tumor volume of nasopharynx) the matched groups in the post-PSM cohort were conducted with the independent t test. Univariate and multivariate analyses of dosimetric parameters were performed with the linear regression model. The variants, which showed an α < 0.1 in the univariate analysis, were enrolled in the multivariate analysis.

IC on the dosimetric parameters of subsequent radiotherapy
For GTVnx, 21 patients and 127 patients were classified as the enlargement group and the control group, respectively. After matching, 20 sub-pairs of 40 patients were generated in the post-PSM cohort. A mean volume enlargement of 20.2% was observed in the enlargement group, and a mean volume reduction of 27.1% was observed in the matched control group. Figure 1 shows a typical case of GTVnx enlargement and its matched case in the control group. Table 1 shows the comparisons of baseline characteristics between the GTVnx enlargement group and the control group in the pre-and post-PSM cohorts. As shown in Table 2, the enlargement group exhibited no significant disadvantages in all of the dosimetric parameters compared with the matched control group, except in the contralateral lens PRV (Dmax, 722 cGy vs. 634 cGy, p = 0.041).

| Impact of GTVnd enlargement after IC on the dosimetric parameters of subsequent radiotherapy
For GTVnd, 44 patients and 144 patients were classified as the enlargement group and the control group, respectively. After matching, 39 sub-pairs of 78 patients were generated in the post-PSM cohort. A mean volume enlargement of 50.6% was observed in the enlargement group, and a mean volume reduction of 40.1% was observed in the matched control group. Figure 2 shows a typical case of GTVnd enlargement and its matched case in the control group. Table 3 shows the comparisons of baseline characteristics between the GTVnd enlargement group and the control group in the pre-and post-PSM cohorts. As shown in Table 4, the enlargement group exhibited no significant disadvantages in all of the dosimetric parameters, compared with the matched control group.

| Univariate and multivariate analyses of dosimetric parameters
To further confirm the association between the tumor volume enlargement and subsequent radiotherapy, univariate and

| DISCUSSION
Due to the complicated anatomical location, the tumor size of NPC has a significant influence on the dosimetric parameters of radiotherapy. [12][13][14] Tumor volume enlargement after IC has been observed in a small proportion of NPC patients despite the high chemotherapy sensitivity of the cancer, 6 but Abbreviations: C, contralateral; D 95 , the minimum dose delivered to 95% of the target; D max , maximum dose; I, ipsilateral; PSM, ropensity score matching; V30 Gy, the relative volume of the structure receiving over 30 Gy; V60 Gy, the relative volume of the structure receiving over 60 Gy.

T A B L E 2 Comparisons of dosimetric
parameters between the GTVnx enlargement group and its matched control group in the post-PSM cohort its influence on the subsequent radiotherapy plan has not yet been investigated. To the best of our knowledge, the current study is the first to address this problem. We compared the dosimetric parameters between patients with tumor volume enlargement and patients with tumor volume reduction after IC, and PSM was adopted to control the balance of other factors, including T stage, N stage, pretreatment GTVnx volume, pretreatment GTVnd volume, and plan type (tomotherapy vs. conventional IMRT). [15][16][17][18][19][20] Our results showed that GTVnx enlargement after IC had no significant impact on most of the dosimetric parameters. This finding is unexpected because it is expected that the primary tumor of NPC is closely related to the dosimetry of PTVs and OARs. A possible explanation for this phenomenon is attributed to the method of delineating the final GTVnx after IC, which was the summation of pre-IC and post-IC GTVnx according the recommendation of the international guidelines. 11 Despite the difference in tumor volume change between the enlargement group and its matched control group, the difference in the final GTVnx was not statistically significant (47.5 cm 3 vs. 40.1 cm 3 , p = 0.484). Additionally, the only disadvantage of the enlargement group was the protection of the contralateral lens PRV (Dmax, 722 cGy vs. 634 cGy, p = 0.041), which would not have significant influence on clinical outcomes because adherence to the dose limit of lens PRV (Dmax <900 cGy) was performed for both groups. Therefore, a GTVnx enlargement of ≥10% after IC has no significant influence on subsequent radiotherapy It is worth mentioning that several studies have investigated the feasibility of using the post-IC GTVnx as the final GTVnx. A randomized controlled study by Yang et al. showed that using the post-IC GTVnx as the final GTVnx did not reduce the local control and survival rate in locally advanced NPC, but the doses to OARs decreased, and the quality of life improved. 21 Another study by Xue et al. also indicated that contouring GTVnx based on the post-IC images achieved satisfactory survival outcome and avoided overdosing of critical neurological structures. 22 Similar results have been reported by several other studies. 23,24 If the post-IC GTVnx is adopted as the final GTVnx in future practice, the potential influence of tumor volume enlargement on subsequent radiotherapy should not be ignored, as the volume of the final GTVnx between the enlargement group and the reduction group would be significantly different.
Our results also showed that GTVnd enlargement after IC had no significant impact on the dosimetric parameters of subsequent radiotherapy. It is noteworthy that the final GTVnd was defined as the post-IC GTVnd only, and there was a significant difference in the final GTVnd between the enlargement group and its matched control group (18.2 cm 3 vs. 8.1 cm 3 , p = 0.017). This insignificant influence of GTVnd enlargement can be attributed to the anatomical location of lymph nodes, which are not adjacent to the majority of the OARs, in most cases. Despite a GTVnd volume enlargement, the dose coverage of PTVs and the protection of OARs can be easily satisfied for most patients with modern radiotherapy techniques, such as the conventional IMRT and tomotherapy. This is supported by the results of the multivariate analysis of dosimetric parameters (Table 5), which indicated that N stage and pretreatment GTVnd volume were not independently associated with the dosimetry of almost all of the OARs. Similar results have also been reported by the study of Yao et al., which analyzed the radiation doses to OARs in 148 NPC patients and showed that N stage was not independently associated with the dosimetry of most OARs. 25 Therefore, a GTVnd enlargement of ≥10% after IC has no significant impact on subsequent radiotherapy.
It should be noted that univariate and multivariate analyses of dosimetric parameters were also performed in the current study. As shown in Table 5, the volume changes of GTVnx and GTVnd after induction chemotherapy (enlargement group vs. control group) were not independently associated with any of the dosimetric parameters of PTVs and OARs, which is consistent with the results discussed above. In addition, the multivariate analysis indicated that T stage, pretreatment GTVnx volume, and plan type were independently associated with the parameters of most OARs, which is in accordance with the results of previous studies. 15,16,[18][19][20] Although PSM was adopted in our study to control the balance between the enlargement group and the control group, it should be noted that there were still some uncontrolled  biases. First, one case in the GTVnx enlargement group and five cases in GTVnd enlargement group were discarded due to the lack of matched case in the control group. Second, some important factors may be not included in the process of matching, such as the distance between the tumor and the OARs, as there was no practical method which can provide such information. Third, cases with larger tumors may be excluded during the matching process, because the pre-IC tumor volumes in the enlargement group were smaller than the control group before matching as shown in Table 1. It is worth mentioning that the larger pre-IC tumor volumes in the control group indicates that larger tumors may be more sensitive to chemotherapy. Similar finding has been reported in the study of Wang et al., which showed that a  T A B L E 5 (Continued) larger tumor volume was independently associated with a higher likelihood of response to induction chemotherapy in head and neck cancer patients. 26 A possible explanation for this phenomenon is that larger tumors are more likely to be involved with abundant blood supply, 27,28 resulting in higher concentration of chemotherapy drugs in the tumor tissues and better treatment responses. 29,30 The results of this study were potentially affected by several factors. First, 10% was adopted as the cut-off value to determine tumor volume enlargement in our study. A higher cutoff value would significantly reduce the number of cases available for the propensity matching (especially for GTVnx) as shown in the supplement Table 1, which depicts the distribution of the relative volume change of GTVnx and GTVnd after IC. Second, the current study did not analyze the potential influence of the chemotherapy regimen, as docetaxel plus cisplatin was the only IC regimen administered at our center. Third, the survival outcomes were not analyzed in our study because the follow-up time (median follow-up time: 21 months) was too short to analyze the survival outcome of non-metastatic NPC, which has a 5-year OS of 70-90%. 4,31 Last, the sample size of our research was small (only 20 pairs of matched patients for GTVnx and 39 pairs of matched patients for GTVnd), which should be taken into consideration while interpreting the results.
To summarize, a tumor volume enlargement of ≥10% in GTVnx or GTVnd after IC has no significant impact on the dosimetric parameters of subsequent radiotherapy in locally advanced NPC.

ETHICS APPROVAL
This study was approved by the Ethics Committee of the Xiangya Hospital of Central South University prior to commencement.