18F-FDG-PET/CT-measured parameters as potential predictors of residual disease after neoadjuvant chemoradiotherapy in patients with esophageal carcinoma

Objective To evaluate the maximum and mean standardized uptake values, together with the metabolic tumor value and the total lesion glycolysis, at the primary tumor site, as determined by 18F-fluorodeoxyglucose positron-emission tomography/computed tomography (18F-FDG-PET/CT), performed before and after neoadjuvant chemoradiotherapy (nCRT), as predictors of residual disease (RD) in patients with esophageal cancer. Materials and Methods The standardized uptake values and the volumetric parameters (metabolic tumor value and total lesion glycolysis) were determined by 18F-FDG-PET/CT to identify RD in 39 patients before and after nCRT for esophageal carcinoma. We used receiver operating characteristic curves to analyze the diagnostic performance of 18F-FDG-PET/CT parameters in the definition of RD. The standard of reference was histopathological analysis of the surgical specimen. Results Eighteen patients (46%) presented RD after nCRT. Statistically significant areas under the curve (approximately 0.72) for predicting RD were obtained for all four of the variables evaluated after nCRT. Considering the presence of visually detectable uptake (higher than the background level) at the primary tumor site after nCRT as a positive result, we achieved a sensitivity of 94% and a specificity of 48% for the detection of RD. Conclusion The use of 18F-FDG-PET/CT can facilitate the detection of RD after nCRT in patients with esophageal cancer.


INTRODUCTION
In locally advanced resectable esophageal cancer, neoadjuvant chemoradiotherapy (nCRT) is currently the standard of care (1,2) . After nCRT, pathological regression is the main prognostic indicator for esophageal cancer (3) . Currently, patients with esophageal cancer undergo surgical resection of the primary lesion after nCRT regardless of the pathological status of the lesion after the therapy. However, for patients with other types of cancer, such as rectal cancer (4) , a watchful waiting approach is often adopted if clinical staging suggests a pathological complete response (pCR) after nCRT. Therefore, if a diagnostic imaging method could accurately identify a pCR after nCRT in patients with esophageal cancer, it would also be possible to identify patients who are the best candidates for a watchful waiting approach, somewhat similar to what is done in cases of rectal cancer (4) , thus avoiding the postoperative complications and high mortality associated with esophagectomy.
Functional imaging, such as 18 F-fluorodeoxyglucose positron-emission tomography/computed tomography ( 18 F-FDG-PET/CT), evaluates metabolic activity and may improve patient selection for further treatment (5) . There have been studies demonstrating the prognostic value of metabolic tumor volume (MTV) and total lesion glycolysis (TLG) for esophageal cancer treated with trimodal therapy (6)(7)(8)(9)(10)(11)(12)(13) . However, the use of the 18 F-FDG-PET/CT to assess the presence of residual disease (RD) after nCRT in patients with esophageal cancer has not been widely studied. In lung and rectal cancer, serial 18 F-FDG-PET/ CT after neoadjuvant therapy can identify predictors of a pathological response (14,15) . However, studies using 18 F-FDG-PET/CT after nCRT in patients with esophageal cancer have produced heterogeneous results. Heneghan et al. (16) evaluated the accuracy of the post-nCRT maximum standardized uptake value (SUV max ) in detecting a pCR and found it to have a sensitivity of only 56%. McLoughlin et al. (17) showed that 18 F-FDG-PET/CT had a sensitivity of 62% and specificity of 44% for the detection of a pCR after nCRT. Gabrielson et al. (18) reported significantly greater SUV changes in nCRT responders than in nCRT nonresponders. However, the authors found no significant difference in SUV between the patients with a pCR and those with a subtotal response. In addition, Elliott et al. (19) evaluated esophageal adenocarcinoma only and showed that 18 F-FDG-PET/CT had poor discriminatory value for clinical application. However, those studies did not evaluate the volumetric parameters determined by 18 F-FDG-PET/ CT and used distinct nCRT regimens. Therefore, our study aims to evaluate the potential of the SUV max , mean SUV (SUV mean ), MTV, and TLG measured at the site of primary esophageal tumor on 18 F-FDG-PET/CT performed before and after nCRT with a platinum-and taxane-based regimen, as well as the change in those values between the two time points and their association with RD.

MATERIALS AND METHODS Patients and study design
This was a retrospective cross-sectional study of patients with esophageal carcinoma, in which we attempted to determine whether the SUV max , SUV mean , and volumetric parameters (MTV and TLG) obtained by 18 F-FDG-PET/CT, before and after nCRT, are associated with the pathological response. We recruited patients from among those under treatment at a single institute between 2009 and 2019. We included 39 patients who had completed a (platinum-and taxane-based) nCRT regimen, followed by esophagectomy with curative intent, and had undergone 18 F-FDG-PET/CT at least twice: before nCRT; and between the end of nCRT and the esophagectomy. The nCRT included the administration of carboplatin or cisplatin concurrent with radiation (41.4, 45.0, or 50.4 cGy).
The patients were staged with endoscopy, CT, and 18 F-FDG-PET/CT, after which they were classified according to the 8th edition of the Union for International Cancer Control staging system (20) . The local research ethics committee approved the study (Reference no. 1492/19).
An experienced pathologist, blinded to the 18 F-FDG-PET/CT findings, evaluated the surgical specimens. A pCR to nCRT was defined as no residual malignant cells detected by hematoxylin and eosin staining in the surgical specimen. We defined RD as the presence of any cancer cells (single cells or cell clusters).

F-FDG-PET/CT acquisition and imaging analyses
A whole-body PET/CT system with time-of-flight capability (Discovery 690; GE Healthcare, Waukesha, WI, USA) was used for the 18 F-FDG-PET/CT acquisition. The patients were instructed to fast for at least six hours and were required to have a blood glucose level ≤ 180 mg/ dL before injection of the 18 FDG (≤ 3.7 MBq/kg of body weight). Image acquisition was initiated approximately 60 min after injection of the radiotracer, and images were acquired from the mid-skull to the mid-thigh. The metabolic activity at the primary tumor site, before and after nCRT, was recorded by using the following parameters: SUV max , SUV mean , TLG, and MTV. Those values were calculated by a single nuclear medicine physician using a radiology workstation (AW VolumeShare 5; GE Healthcare). The SUV thresholds used in order to define the boundaries of the lesions were established by visual analysis. The total volume of interest that circumscribed the metabolic area was calculated automatically by the dedicated software.

Statistical analysis
Receiver operating characteristic (ROC) curves were used in order to establish the sensitivity and specificity of the distinct operating points of the four parameters measured before and after nCRT, as well as the differences between the pre-and post-nCRT values, thus allowing the post-nCRT status of the primary tumor site to be classified as a pCR or RD. The result of the histopathological analysis of the surgical specimen was used as the gold standard. The areas under the curve (AUCs) and the corresponding 95% confidence intervals (95% CIs) were used in order to evaluate the accuracy of the four parameters in that classification. Finally, we estimated the sensitivity and specificity of those parameters for the identification of RD when no cancer cells were detected through visual inspection of the primary tumor site after nCRT. Any residual uptake above the background level at the primary tumor site was used as the threshold.
Descriptive statistics-including means and standard deviations; medians and ranges; and absolute and relative frequencies-are presented for some variables. We also present the sensitivity and specificity of the best operation point to classify the patients as achieving a pCR or having RD. Statistical analyses were performed with the Stata software package, version 16.1 (StataCorp, College Station, TX, USA). The level of statistical significance was set at p < 0.05.

Patient baseline characteristics
Our analysis included 39 patients who underwent 18 F-FDG-PET/CT before nCRT with a platinum-and taxanebased regimen and between the nCRT and the esophagectomy. The mean time from the pre-treatment 18 F-FDG-PET/CT to the beginning of the nCRT was 12 ± 6 weeks. The mean time from the last cycle of the nCRT to the post-treatment 18 F-FDG-PET/CT was 9 ± 4 weeks, and the mean time from the end of nCRT to surgery was 16 ± 6 weeks.

Pathological response
The ROC curve analyses of the variables determined by 18 F-FDG-PET/CT, in terms of their ability to predict RD, are shown in Table 2, as well as in Figures 1 and 2. The pre-treatment values for the 18 F-FDG-PET/CT parameters were not found to be predictors of RD. For the post-treatment 18 F-FDG-PET/CT parameters, the AUCs were similar among all four of the variables related to the primary tumor and those AUCs were statistically significant ( Figure 1). After nCRT, the AUC was 0.7169 (95% CI: 0.5541-0.8797) for the SUV max , 0.7169 (95% CI: 0.5537-0.8801) for the MTV, 0.7196 (95% CI: 0.5544-0.8848) for the SUV mean , and 0.709 (95% CI: 0.5449-0.8731) for the TLG. For the difference between the pre-and post-treatment values of the four 18 F-FDG-PET/ CT parameters, only the SUV parameters (SUV max and SUV mean ) showed significant ability to detect RD ( Figure  2). Table 2 summarizes those findings.
Given that the main utility of 18 F-FDG-PET/CT in esophageal cancer is to detect all patients with RD, which improves the selection of surgical candidates, the sensitivity of the modality is more important than is its specificity. Therefore, if the SUV or volumetric parameters of the primary tumor on 18 F-FDG-PET/CT after nCRT were not "zero" (i.e., the nuclear medicine physician visually detected a region of uptake above the background level at the primary tumor site), the sensitivity for predicting RD was high (94.4%; 95% CI: 72.7-99.9%), although the specificity was intermediate (47.6%; 95% CI: 25.7-70.2%). Figure 3 depicts a patient with residual uptake at the primary tumor site after nCRT. Figure 4 depicts another patient, in whom there was no visually detectable  * In accordance with the 8th edition of the Union for International Cancer Control staging system (20) .   (Table 3).

DISCUSSION
In this study of patients with esophageal cancer, we have shown that the 18 F-FDG-PET/CT parameters obtained for the primary tumor site after nCRT has the potential to predict RD. In this context, the presence of any visually detectable uptake above the background level at the site of the tumor seems to be the best predictor of RD. Our results suggest that some patients without visually detectable uptake on 18 F-FDG-PET/CT after nCRT would benefit from a watchful waiting approach, somewhat similar to that applied in cases of rectal cancer (4) , which avoids the risk of postoperative complications and mortality associated with esophagectomy. Such patients accounted for approximately 28% of our sample. If a watchful waiting approach had been adopted in those cases, the overall mortality rate could have been lower.
Our study has some limitations. First, it was a retrospective, single-center study with small sample size. In addition, although our results provide evidence to support the use of 18 F-FDG-PET/CT to predict RD after nCRT, the criterion used for 18 F-FDG-PET/CT classification as positive (any uptake above the background level) is not perfect and failed to detect RD in one of the 18 patients with RD in our sample. Therefore, the method should be used with extreme caution for surgical decision-making, and every choice should be shared with the patient and their family. As a general rule, patients should still be referred for resection. Nevertheless, 18 F-FDG-PET/CT could be useful in some cases in which surgery is indicated. For example, in patients who deteriorate in the setting of neoadjuvant therapy, who are at higher risk of complications of esophagectomy, and in whom the 18 F-FDG-PET/CT variables favor the attainment of a pCR, the medical team could offer, in consultation with the patient and their family, the option of adopting a watchful waiting approach.
Despite the fact that the SUV and volumetric parameters are known to be associated with the long-term prognosis in esophageal cancer (6-13,21) , there are divergent results in the literature regarding the utility of 18 F-FDG-PET/CT in predicting the pathological response to nCRT (11,22,23) . Most of the relevant studies have used a variety of neoadjuvant regimens, have not assessed th 18 F-FDG-PET/CT parameters after nCRT, and have not taken a standardized approach to data analysis. Arnett et al. (22) showed that the 18 F-FDG-PET/CT parameters (SUV max , SUV max normalized to liver uptake, and SUV max normalized to blood pool uptake) measured pre-and post-nCRT were not significantly associated with a pCR in a sample of 193 patients, most of whom had adenocarcinoma. However, the authors did not use ROC curves to define a metabolic threshold to separate patients with a pCR from those with RD. In addition, most of the patients in our sample had squamous cell There was also nonspecific diffuse uptake in the right masseter and lateral pterygoid muscles, which could be explained by muscle contraction, given that there were no evident anatomical alterations. carcinoma, which is known to have a higher probability of a pathological response after chemoradiotherapy, as demonstrated elsewhere (24) . That could explain the superiority of our results. In a recent study, Choi et al. (25) , analyzing a cohort of patients undergoing trimodal therapy, showed that pre-and post-treatment volumetric parameters from 18 F-FDG-PET/CT are independent variables associated with a pCR. However, those authors evaluated 18 F-FDG-PET/CT parameters only as prognostic variables and did not propose the use of 18 F-FDG-PET/CT as a diagnostic tool for detecting RD. Currently, clinicians lack a reliable tool to facilitate the decision between surgery and a watchful waiting approach after nCRT. Our study provides evidence that 18 F-FDG-PET/CT could be a useful tool for the detection of RD and has high sensitivity when any residual uptake above the background level at the primary tumor site is used as a threshold.
It should be borne in mind that an inflammatory reaction due to radiation exposure may partially explain the low specificity of 18 F-FDG uptake in the definition of RD. Previous studies have suggested that an inflammatory response could have a confounding effect on the 18 F-FDG-PET/CT variables after neoadjuvant therapy (26,27) . Therefore, the interval between the end of the nCRT and the post-treatment 18 F-FDG-PET/CT may be a key factor that could explain some of the variability of the results among the studies and could contribute to the limited specificity.
In the present study, the mean time from the last cycle of the nCRT to the post-treatment 18 F-FDG-PET/CT was relatively short (nine weeks). It is possible that a longer interval would have increased the specificity. Therefore, future studies should attempt to determine whether a longer interval between the last nCRT cycle and 18 F-FDG-PET/ CT could improve the accuracy of the examination in the identification of RD.
It is noteworthy that, in our study, any residual uptake above the background level at the primary tumor site was found to be the best threshold to classify the 18 F-FDG-PET/CT study as positive or negative for RD. Therefore, the quantitative and semiquantitative parameters (SUV max , SUV mean , MTV, and TLG) do not appear to be fundamental for the purpose of defining the pathological response, although that should be better analyzed in future studies.
Combining the 18 F-FDG-PET/CT data with those obtained by other diagnostic methods or with clinical and demographic data could increase the accuracy for the identification of a pCR. In a study of esophageal squamous cell cancer conducted by Molena et al. (28) , a ≥ 70% reduction in the SUV max combined with a normal appearance on endoscopy and a lack of RD on biopsy was found to increase the chance of achieving a pCR. Zhang et al. (29) analyzed a model to predict a pCR that combines 18 F-FDG-PET/CT, clinical data, and demographic features. They found that the model accurately predicted a pCR. Therefore, future analyses of our data are necessary to determine whether combining the results of the 18 F-FDG-PET/CT study with clinical and demographic parameters, as well as with the results of other diagnostic tests, could further improve the accuracy for the detection of pCR.
The results of the present study do not provide a definitive answer for clinicians who manage esophageal cancer and should be interpreted in the context of the aforementioned limitations. Therefore, larger, controlled prospective studies are warranted in order to determine the true accuracy of 18 F-FDG-PET/CT for the detection of RD.

CONCLUSION
The use of 18 F-FDG-PET/CT after neoadjuvant therapy in patients with esophageal carcinoma has the potential to predict the pathological response. The parameters measured by 18 F-FDG-PET/CT also facilitate the selection of patients who are eligible for a watchful waiting approach.