The Predictive Value of Tumor Volume, Inammatory Biomarkers, and Their Dynamic Changes on Early Tumor Response for Elderly Patients With Esophageal Squamous Cell Carcinoma Underwent Radiotherapy

Background: To investigate the tumor volume, pre-treatment inammatory biomarkers (pre-IBs), and their dynamic changes on early tumor response (ETR) in elderly patients ( ≥ 70 years) with esophageal squamous cell carcinoma (ESCC) underwent radiotherapy. Methods: The ETR was assessed according to RECIST 1.1 at 1 month after radiotherapy. The tumor volume ((gross tumor volume (GTV) at the initial treatment planning (GTVi), and GTV at shrinking irradiation eld planning (GTVs)), IBs (neutrophil/lymphocyte (NLR), platelet/lymphocyte (PLR), and lymphocyte/monocyte (LMR)) which also included during treatment IBs (dur-IBs), and clinical variables were collected and analyzed from 197 patients received radiotherapy at our institution between 2015 and 2020. The tumor volume change rate (TVCR) and dynamic changes of IBs (delta-IBs) were dened as follows: TVCR=(1-GTVs/GTVi)×100%, delta-IBs=1-dur-IBs/pre-IBs. A nomogram based on logistic regression analysis were then established for predicting ETR. Results: GTVi and pre-LMR signicantly decreased, pre-NLR, and pre-PLR signicantly increased during radiotherapy or chemoradiotherapy (all P<0.001). Multivariate analysis indicated that TVCR [OR, 0.197; 95%CI, 0.093-0.414; P<0.001], pre-NLR [OR, 2.568; 95%CI, 1.031-6.394; P=0.043], and delta-NLR [OR, 2.831; 95%CI, 1.126-7.119; P=0.027] were statistically signicant with ETR. And c-index of the nomogram established by combining all independent predictors for ETR was 0.769 [95%CI, 0.161–0.302]. Conclusion: TVCR, ESCC on in that were signicantly associated with ETR in ESCC patients.

the social security system, the elderly patients have accounted for 50% of the EC cases [2][3][4] . And most elderly patients are accompanied by other basic diseases, poor physical and physiological conditions, and sometimes show accompanied by malnutrition, electrolyte disorder, and dysfunction of the heart, liver, and kidney, which are not considered suitable for operation [5] . So, radiotherapy, or chemoradiotherapy (CRT) become the main treatment for elderly EC patients, and it is reported that about 70% of EC patients need radiotherapy at different stages of tumor treatment [6] . However, local recurrence and distant metastasis are the main causes of treatment failure, and the risk of perforation and bleeding caused by radiotherapy in the short term after recurrence is high, and the effect of second-line chemotherapy is poor [7,8] . Even if the elderly EC patients have the same clinical stage, and they are also given the same treatment as young patients, but the curative effect could be very different [9] .
So, it is necessary to predict the early tumor response (ETR) of radiotherapy for elderly EC patients to optimize the individual treatment strategy, increase radiosensitizer appropriately, and follow-up promptly to improve the local control rate and the survival rate.
The earlier effective treatment is detected, the greater the bene t to elderly patients in clinical practice.
Tumor volume is the most intuitive manifestation of the therapeutic effect, and with the improvement of radiotherapy technology, we can routinely evaluate tumor volume during radiotherapy. Some studies have also reported that tumor volume is a potential predictor for EC patients [10][11][12] . Moreover, the interaction between in ammatory cells and tumor cells can change the dynamic balance of tumor tissue to in uence the therapeutic effect [13,14] . Now, more and more studies investigate the relationship between in ammatory biomarkers (IBs) and the prognosis of patients with EC [15][16][17][18] . Additionally, cancer-related in ammation is a key determinant of disease progression, and patient survival, especially the host response in the form of systemic in ammation has been shown to independently predict prognosis [19][20][21] .
Whether tumor volume or IBs is constantly changing during treatment, the tumor volume change rate (TVCR) and dynamic changes of IBs (delta-IBs) may be more valuable in re ecting the real-time response to radiotherapy and be more accurate for predicting ETR. Most of these prognostic biomarkers in previous studies were collected at the pre-treatment and did not involve the response of the tumor to radiotherapy during treatment [10-12,15−18] . Meanwhile, elderly patients are often excluded or at least underrepresented in randomized trials. Now, we must recognize that elderly cancer patients do not necessarily have a short life as life expectancy keeps on increasing. They could be adjusted treatment strategy and monitored if the early tumor response could be predicted. Hence, this study aimed to investigate tumor volume, IBs, and their dynamic changes on ETR in elderly ESCC patients (≥ 70 years) who underwent radiotherapy.

Page 4/20
We performed a retrospective analysis of 197 ESCC patients aged ≥ 70 years who were underwent radiotherapy with or without chemotherapy in Shandong Cancer Hospital between 12/2015 and 02/2020. Inclusion criteria include the following: (1) new diagnosed ESCC and con rmed by histopathology; (2) receipt of 50-64Gy radiotherapy with or without chemotherapy, and a repeated computed tomography scan was performed before and follow 40Gy for shrinking irradiation eld; (3) availability of complete blood cell counts (CBCs), and radiology date (include esophagography, enhanced computed tomography, radiotherapy planning evaluation system and imaging data) before and during treatment; (4) the local focus could not be resected; (5) Karnofsky performance status ≥ 70; (6) no history of malignant tumors, no intolerable serious medical diseases, no distant metastasis, no active esophageal bleeding; (7) no autoimmune, blood, and infectious diseases or fever. All patients in this study were restaged according to the American Joint Committee of Cancer 7th edition TNM classi cation and staging system. For the clinical and treatment characteristics of the patients see Table 1.

Treatment characteristics
All patients were underwent 3-dimensional conformal radiotherapy or intensity-modulated radiotherapy.
Radiotherapy alone or concurrent chemoradiotherapy was performed according to the clinical stage and the physical condition of patients. All radiotherapy plans were generated in the Eclipse system (Varian Medical Systems, Palo Alto, CA, Version 13.5.35), and delivered with 6 MV photons beams. The prescribed doses of radiotherapy were 50-64Gy at 2.0Gy per fraction once daily and ve fractions per week. Plans were normalized to 95% of the plan tumor volume received 100% of the prescribed dose. The dose of all organs at risk (OARs) was controlled below the safe range. The chemotherapy regimen has mainly consisted of two platinum-based regimens. One was cisplatin with uorouracil, one was cisplatin with paclitaxel.
The delineation of target volumes, OARs, and doses of chemotherapy regimens followed the guidelines of the Chinese Society of Clinical Oncology and the National Comprehensive Cancer Network for EC.

Response evaluation
All patients performed esophagography and enhanced computed tomography before and at 1 month after radiotherapy. The imaging data were analyzed and the ETR of all patients were assessed by an imaging deputy chief physician and a radiotherapy deputy chief physician according to Response Evaluation Criteria in Solid Tumor 1.1 (RECIST 1.1) without knowledge of the results of TVCR and IBs studies. Patients with an outcome of complete response (CR) or partial response (PR) were subsequently classi ed as responders, while those who had an outcome of stable disease (SD) or progressive disease (PD) were de ned as non-responders.
TVCR and delta-IBs calculation IBs included neutrophil/lymphocyte (NLR), platelet/lymphocyte (PLR), and lymphocyte/monocyte (LMR) in this study. The CBCs were collected at different time points. The rst time point was before the start of radiotherapy which we de ned as pre-IBs. The second was when preparing shrinking irradiation eld planning which we de ned as dur-IBs. The gross tumor volume (GTV) de ned on radiotherapy planning refers to the range of tumor lesions with certain shapes and sizes displayed by existing auxiliary examination methods including computed tomography and esophagography, and its stereoscopic imaging re ects the actual shape of the tumor. And we de ned GTV including the primary tumor and involved lymph nodes at initial treatment planning as GTVi, at shrinking irradiation eld planning as GTVs. All GTVi and GTVs were extracted from the Varian treatment planning system.

Statistical analysis
Converting continuous variables into binary variables by using receiver operating characteristic (ROC).
The difference between pre-treatment and during treatment of tumor volume and IBs were compared using Wilcoxon's sign rank test. Univariate logistic regression analysis was performed to estimate the odds ratio (OR) and con dence interval (CI) to evaluate the effect of independent variables on ETR. To avoid omitting indicators that might be of clinical signi cance, factors that had P < 0.1 in univariate analyses were subjected to multivariate analysis. Scatter plot and Spearman's rank correlation coe cient (r) values were performed in the same kind of parameters. A signi cant difference was considered if a Pvalue was < 0.05 in 2-sided. And Statistical evaluation was conducted with IBM SPSS Statistics 25.0 software (SPSS Inc., Chicago, IL). A nomogram for possible prognostic factors associated with ETR was established by R 3.4.4 software (Institute for Statistics and Mathematics, Vienna, Austria), and the predictive accuracy was evaluated by the concordance index (c-index).

Nomogram for predicting prognosis of ETR
The nomogram was established by multivariate logistic regression according to all signi cantly independent factors for ETR in elderly ESCC patients (Fig. 4a). And Fig. 4a also shown that the TVCR had the best predictive value for ETR, then pre-NLR, delta-NLR in this study. The calibration curve of the nomogram for ETR demonstrated good agreement between nomogram prediction and actual observation (Fig. 4b). And the C-index for the prediction nomogram was 0.769 (95%CI, 0.698-0.839) by internal bootstrapping validation.

Discussion
Nowadays, there is still no established standardized treatment strategy for elderly EC patients due to their special characteristics. Although considerations which include age, functional status, risk of treatmentrelated morbidities, life expectancy, and personal preference should be taken into account when making treatment decisions, the prognosis of elderly patients with EC is still poorer than young non-elderly patients [5] . Furthermore, RECIST 1.1 which has been widely used in the clinic to evaluate tumor's response to anticancer therapy still has some limitations. First, its main evaluation criteria are to measure changes in the longest diameter of assessable lesions, while ignoring short diameter and tumor volume; secondly, the one-dimensional measurement based on morphological changes ignores molecular changes and cannot effectively re ect the biological changes in tumors, which may lead to inaccurate prediction of treatment responses; thirdly, the evaluation of e cacy by RECIST 1.1 criteria is usually carried out one or three months after treatment, which may delay the detection of disease progression and recurrence [22] . Hence, it is necessary to explore new biomarkers to predict the ETR of elderly ESCC patients. Tumor volume can be easily gotten from the Varian treatment planning system, and IBs have the advantages of economy, simple detection, and easy to be accepted by the majority of patients. Based on those, we investigated the tumor volume, pre-IBs, and their dynamic changes on ETR in newly diagnosed elderly ESCC patients (≥ 70 years) who underwent radiotherapy. And our present study found that TVCR, pre-NLR, and delta-NLR were signi cantly associated with ETR in elderly ESCC patients. Furthermore, the developed nomogram based on all independent predictors had a better prediction ability for ETR.
Generally speaking, the bigger tumor volume is, the great tumor burden is. As far as EC is concerned, Boggs et al thought that the greater the tumor volume load, the lower the survival time of patients, and tumor volume is an independent prognostic factor affecting the survival of patients [10] . Chen et al retrospectively analyzed the clinical data of 187 patients after radiotherapy and found that the survival time of patients with higher GTV (> 39.41cm 3 ) which was based on the radiation therapy planning system was signi cantly lower than that of the comparison group (GTV < 39.41cm 3 ) [11] . Furthermore, Créhange et al have also retrospectively analyzed the tumor volume of 148 patients with EC underwent radiotherapy and found that overall survival (OS) of patients with tumor volume ≥ 100cm 3 which was calculated by assimilation as the sum of two pairs of truncated cones was signi cantly lower than that of patients with tumor volume < 100cm 3 (P = 0.041) [12] . The tumor volume in the present study was automatically calculated by the radiotherapy planning system, and neither GTVi nor GTVs was associated with ETR. One reason was that all the ESCC patients were elderly which had dull sensation, late appearance of self-feeling symptoms, and late medical treatment so that most of the tumor volume were larger than non-elderly patients when diagnosed. Another was that esophagus was a hollow organ, the hollow tissue would affect the accuracy of the automatic drawing of tumor tissue. However, it did not mean that tumor volume had no predictive value in elderly ESCC patients. It seemed to be more valuable to quantify tumor volume change during treatment from the aspect of disease regression. Jobbour et al had reported that tumor volume reduction during CRT could predict post-CRT survival in NSCLC [23] . Yang et al also found that volume reduction rate was an outcome predictor for head-and-neck cancer patients treated with radiotherapy [24] . Interestingly, NSCLC patients with pronounced volume regression had worse locoregional tumor control and overall survival in Brinkv et al study [25] . However, elderly ESCC patients who were bigger TVCR (≥ 6.24%) had better ETR in this study which might suggest that TVCR might be a more sensitive indicator than tumor volume. Moreover, there was a signi cant decrease between GTVi and GTVs which indicated that radiotherapy was an effective treatment for elderly ESCC patients. Currently, radiotherapy physicians still manually delineate and calculate tumor volumes by computer systems and therapeutic planning systems, while many factors such as doctors' clinical knowledge, experience, energy, and status determine that there are some differences in drawing quality between different doctors and different patients [26] . There is no doubt that these also affect target delineation and calculation of tumor volume. It is possible to solve these problems with arti cial intelligence (AI) at the age of precision medicine. Lin et al have used AI technology to automatically draw nasopharyngeal tumors on magnetic resonance images, which provided a solution for accurate and e cient delineation of radiotherapy targets [27] . Choi et al have con rmed the plausibility of deep learning-based automatic segmentation for clinical implementations in breast cancer [28] . So, the prospect of precision medicine based on radiotherapy for tumors will be promising with the development of AI.
Apart from tumor volume, some studies also found that in ammatory response is closely associated with tumor prognosis [19,29] . The in ammatory response can participate in and promote the malignant transformation, migration, and diffusion of tumor cells by up-regulating cytokines, producing in ammatory mediators, inhibiting apoptosis, promoting angiogenesis, and inducing DNA mutation [14,30−31] . Lymphocytes, neutrophils, platelets, and monocytes have always been considered to play an important role in immune balance and in ammatory response. Lymphocytes are considered to be tumor suppressor cells, which can induce tumor cell death and inhibit tumor progression [32] . An abnormal increase of neutrophils may lead to tumor proliferation, increased potential of local invasion and metastasis, tumor vascularization, and help tumor cells escape immune surveillance [33] . Platelets can cooperatively activate tumor cells and are essential for early metastasis of cancer [34] . Monocytes may be involved in tumor growth and promote tumor progression and metastasis [35] . Furthermore, the complex IBs which include NLR, PLR, and LMR can re ect the dynamic relationship between anti-tumor and promoting tumor. Our present study found there was a signi cant increase in pre-NLR during radiotherapy, as well as in pre-PLR, and a signi cant decrease in pre-LMR. This association between changes of IBs and radiotherapy at different time points could at least be partly explained. Besides immunity gradually decreases with age, larger tumor volume (P < 0.0001), concurrent chemotherapy (P < 0.0001), and stage III disease (P = 0.05) would lead to a lower count of lymphocytes [36] .
The predictive value of NLR in EC has been con rmed in many studies. Yoo et al have conducted a retrospective study on 138 locally advanced EC patients who received concurrent CRT, and found that high pre-NLR group the progression-free survival (PFS) time and OS time were signi cantly shorter in the high pre-NLR group (NLR ≥ 2.0) were signi cantly associated with decreased PFS and OS (all P < 0.05) [15] . Zhou et al have also analyzed 517 locally advanced EC patients who received CRT and pointed out that pre-NLR > 5 was an independent predictor of PFS and overall OS (all P < 0.001). Similar to these results, pre-NLR was an independent predictor for ETR in elder ESCC patients who underwent radiotherapy in this study [16] . Additionally, delta-NLR was also signi cant associated with ETR, and elderly ESCC patients who were with low delta-NLR (<-1.728) would get better ETR which might suggest that delta-IBs were also predictors for them. Choil et al have found that changes in NLR after neoadjuvant treatment showed a statistically signi cant correlation with advanced breast cancer patients' survival [37] . Templeton et al reported that early decline of NLR was associated with favorable outcomes, whereas an increase was associated with worse outcomes for metastatic renal cell carcinoma with targeted therapy [38] . Whether pre-NLR or delta-NLR, these studies did not speci cally analyze the elderly patients. This was the rst signi cant study that analyzed the predictive value of pre-NLR and delta-NLR in elderly ESCC patients. In therapy, delta-NLR should be better re ect the sensitivity of the tumor to treatment. However, pre-NLR had a better predictive value than delta-NLR in this study which might be that pre-NLR was better able to re ect the state of the body's original response.
Recently, many scholars have studied the prognostic value of PLR and LMR in cancer patients, but their role in ESCC is still controversial. Xie et al found that PLR had the predictive ability for ESCC patients with different stages, which was an independent predictor for stage I ~ II, but not for stage III ~ IV [17] . While Jun et al thought that there was no signi cant correlation between PLR and postoperative survival time by analyzing the clinical data of patients with ESCC [39] . Another study also reported that a low LMR (< 4) was a signi cant and independent predictor of poor survival in non-elderly EC patients who received curative thoracoscopic esophagectomy [18] . But our present study found that PLR and LMR were not signi cant with ETR in multivariate analyses for elderly ESCC patients. The reason might be that radiotherapy might result in part of necrosis and death of cancer cells as well as the surrounding tissues, which in turn cause an in ammatory response similar to the wound healing response. Interestingly, there was a correlation between delta-NLR and delta-LMR, as well as delta-PLR, while only delta-NLR was signi cantly associated with ETR in this study.
We have con rmed that the IBs and TVCR were signi cant with early tumor response in ESCC patients who underwent radiotherapy or CRT by different retrospective studies [40][41] . However, we did not analyze the tumor volume and IBs on ETR in elderly ESCC patients. To our knowledge and limited literature searches, this was the rst report that describes the prognostic signi cance of ETR combined with tumor volume, IBs, and their dynamic changes in elderly ESCC patients who underwent radiotherapy. But several limitations would be addressed here. First, this was a retrospective, single-center, and small population study that was entire of elderly ESCC patients, which might limit the universality of the results. Second, oncology does not have a speci c age threshold for elderly patients with cancer. And we de ned the elderly as the patient whose age ≥ 70 years. Last, there is no consensus on the exact cut-off value for IBs, and various methods including ROC and median values have been used to determine the optimal segregation points. Consequently, larger, multicenter, and prospectively designed clinical trials are required to con rm these initial results.

Conclusions
In summary, our study con rmed that TVCR, pre-NLR, and delta-NLR before and during RT were associated with ETR in elderly ESCC patients. Moreover, TVCR had the best predictive value, suggesting that pre-NLR may be an important factor predicting the ETR. And the developed nomogram with superior prediction ability for ETR could assist in patients counseling and guide to make individual treatments and follow-up strategies for elderly ESCC patients.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
The ethics committee of Shandong Cancer Hospital and Institute approved the study. Also, administrative permissions to access the patients' clinical data used in our research were granted by the ethical review board. And written informed consent was waived because of its retrospective design.

Figure 2
The forest plot of the multivariate analysis of clinical, TVCR, and in ammatory parameters in predicting the early tumor response in elderly patients with ESCC. Prediction Nomogram for early tumor response (ETR). (a) Nomogram predicts ETR prediction based on TVCR, pre-NLR, and delta-NLR in elderly patients with ESCC. The nomogram is used by totaling the points identi ed at the top of the scale for each independent factor. This total point score is then identi ed on the total points scale to determine the probability of risk prediction. (b) The calibration curve of the nomogram for ETR, and the c-index for ETR prediction was 0.769.