Impact of dose reduction and treatment delay of neoadjuvant chemotherapy in gastric and esophageal adenocarcinoma

Background Neoadjuvant chemotherapy in resectable gastric and esophageal adenocarcinoma is often hampered by toxicities and fragile patients resulting in dose reductions and/or treatment delays. The aim of this study was to assess how these treatment modications affect outcome. Methods A series of 63 consecutive patients treated 2008-2014 with neoadjuvant EOX (epirubicin, oxaliplatin and capecitabine) and surgical resection, with or without adjuvant treatment, were reviewed. Chemotherapy dose index (DI), i.e. the ratio of actual to planned cumulative dose, and time index (TI), i.e. the ratio of planned to actual total duration, were calculated. Associations of neoadjuvant EOX DI and TI with histopathologic response were analysed with binary logistic regression. Time to recurrence (TTR) and overall survival (OS) were estimated using Kaplan-Meier analyses. Results Statistically signicant associations were found between neoadjuvant EOX TI >= 0.95 and a major histopathologic response (0-10% residual cancer cells) and between neoadjuvant EOX DI >= 0.95 and a response with 0-50% residual cancer cells. Signicantly improved TTR and OS were seen in patients with a major histopathologic response. Conclusions Our results suggest that treatment delays of neoadjuvant chemotherapy in gastric or esophageal adenocarcinoma should be avoided in order to achieve a major response.

improved TTR and OS were seen in patients with a major histopathologic response. Conclusions Our results suggest that treatment delays of neoadjuvant chemotherapy in gastric or esophageal adenocarcinoma should be avoided in order to achieve a major response.

Background
For patients with resectable gastric or esophageal adenocarcinoma perioperative (i.e. neoadjuvant + adjuvant) chemotherapy is standard of care in many countries, particularly in Europe. This approach is based on the MAGIC trial [1] and the FFCD 9703 trial [2], where the addition of chemotherapy resulted in a 13-14% improved 5-year survival compared to surgery alone. In the MAGIC trial the ECF regimen (epirubicin, cisplatin and uorouracil) was used and the FFCD 9703 trial had a similar regimen but without epirubicin. Since the REAL-2 trial [3] in advanced gastric and esophageal cancer demonstrated that the more convenient EOX regimen (epirubicin, oxaliplatin and capecitabine) yielded a longer survival compared to ECF, most centers in Sweden chose to use EOX also in the perioperative setting, with three cycles before and three cycles after surgery.
Delivering triplet chemotherapy in gastric and esophageal cancer is a major challenge due to toxicities and fragile patients, commonly leading to dose reductions and treatment delays or even premature discontinuation, but little is known about how these treatment modi cations affect outcome.
A widely used method to describe treatment delivery is the Hryniuk model [4] of relative dose intensity (RDI), which is the ratio of actual to planned dose intensity, where dose intensity is the cumulative dose divided by the total treatment duration. There are several publications on the relationship between RDI and survival in various malignancies but few have assessed the differential contribution of the individual components of RDI.
In gastric cancer a few studies have investigated the relationship between RDI and survival in palliative [5] and adjuvant [6] settings, but to our knowledge there are no previous reports on associations between RDI (or its individual components) of neoadjuvant chemotherapy and outcome.
The primary aim of this study was to assess whether dose reductions and/or treatment delays affect histopathologic response, in a consecutive cohort of patients treated with neoadjuvant EOX for resectable gastric or esophageal adenocarcinoma.

Study design and participants
From a cohort previously used in studies on prognostic and predictive biomarkers in resectable gastric and esophageal adenocarcinoma [7,8], we identi ed 99 patients who started neoadjuvant EOX at the Skåne University Hospital in Lund and Malmö between January 1, 2008 and December 31, 2014. Out of these there were 63 patients who after neoadjuvant EOX proceeded to surgical resection, and where detailed chemotherapy data was available. See Figure 1 for an overview of the cohort and Table 1

Histopathologic response
The extent of residual cancer cells (0%, 1-10%, 11-50% or > 50%) in the primary tumor site of the resected specimens were histologically assessed (blinded to treatment data) using the tumor regression grading system described by Chirieac et al. [10].

Statistical analysis
Based on the distributions of EOX DI, TI and RDI, various cutoffs in steps of 0.05 or 0.10 were used in binary logistic regression to calculate odds ratios (OR) for dichotomized histopathologic response. Since no other factor than DI and TI, respectively, had a statistically signi cant OR and because of the limited sample size no adjusted analyses were performed. Receiver operating characteristic (ROC) curves were also applied for DI and TI but did not add any information of additional value. Differences in patient and clinical characteristics between dichotomized groups of neoadjuvant DI and TI were assessed using chisquare test for categorical variables and Mann-Whitney U test for continuous variables. Time to recurrence (TTR) was de ned as time from diagnosis to time to biopsy or radiology proven recurrence (excluding death) and overall survival (OS) was de ned as time from diagnosis to date of death (regardless of cause). Differences in Kaplan-Meier survival curves were estimated using log-rank test. Follow-up time was calculated using inverse Kaplan-Meier estimation. All statistical tests were 2-sided and a p-value < 0.05 was considered statistically signi cant. IBM ® SPSS ® Statistics version 25.0.0.2 for Mac was used for all statistical analyses.

Distributions of neoadjuvant EOX DI, TI and RDI
In the cohort of 63 patients that proceeded to surgical resection the distributions of neoadjuvant EOX DI, TI and RDI were heavily left-skewed as depicted in Figure 2b. Median value for DI was 0.93, for TI 1.00 and for RDI 0.89. All patients but one had a DI less than 1.00 due to rounding down of the capecitabine dose to the nearest multiple of tablets (500 mg and 150 mg). For the majority (60.3%) of the patients TI was 1.00, i.e. the treatments were delivered without delays.
Neoadjuvant EOX DI and TI vs. histopathologic response As shown in Table 2, the only factor with a signi cant OR for a major response (0-10% residual cancer cells) was TI ³ 0.95 with OR 8.40. For response with 0-50% residual cancer cells the only factor with a signi cant OR was DI ³ 0.95 with OR 3.14. The distributions of four-tiered histopathologic response across dichotomized DI and TI are shown in Figure 3. There were no signi cant differences in patient or clinical characteristics between dichotomized groups of DI and TI at cutoff 0.95 (Additional le 1: Table   S1).
TTR and OS in relation to histopathologic response TTR and OS were signi cantly improved in patients with better histopathologic response (Figure 4), with the largest difference noted with the cutoff at 10% residual cancer cells. Patients with a major response had a 5-year OS of 87% compared to 47% for patients with more than 10% residual cancer cells.

Discussion
The main nding of the present study is that tight adherence (TI ³ 0.95) to the chemotherapy time schedule was associated with a higher probability of a major histopathologic response (0-10% residual cancer cells) in patients receiving neoadjuvant EOX for resectable gastric or esophageal adenocarcinoma. We also found a signi cant association between a high cumulative dose (DI ³ 0.95) and a more modest response (0-50% residual cancer cells). Since the aim of neoadjuvant chemotherapy is to reduce the tumor burden before surgery, the goal should be to achieve as good a response as possible. Therefore, an interpretation of our results could be that it is more important to avoid treatment delays than to give full dose in this context. The TI cutoff at 0.95 implies that the total treatment duration should not be delayed by more than a factor 1/0.95 » 1.05, i.e. not more than one day per 21-day cycle.
Previous studies [11][12][13][14][15][16] in this treatment setting have proposed small tumor size (but not T stage), differentiated tumor and Laurén intestinal subtype as predictors for histopathologic response to chemotherapy, but there are con icting results regarding Laurén subtype [17]. We found no other factors than TI and DI to be associated with histopathologic response, though we did not have data on tumor size.
In the present study we chose to use histopathologic response, rather than survival, as primary endpoint. The reason for this is that response is usually a direct effect of the given treatment, whereas survival may be in uenced by several other factors, such as age, comorbidities, tumor stage, surgical quality and complications. Therefore, the neoadjuvant setting, where chemotherapy is followed by surgical resection with histopathologic response evaluation, is a very good model for studies on chemotherapy dose intensity.
Regarding the clinical signi cance of histopathologic response, the literature is con icting, as some studies have found it to be an independent predictor for survival [18] and others have not [19]. The present study showed signi cant associations between survival and histopathologic response, especially in patients with 0-10% residual cancer cells who had an excellent survival. It should however be pointed out that these were univariable analyses, not taking other prognostic factors into account.
In the present investigation we applied a similar methodological approach as in a study by Nakayama et al. [9] on patients with metastatic colorectal cancer, evaluating irinotecan-based (FOLFIRI regimen) and oxaliplatin-based (FOLFOX regimen) chemotherapy, wherein it was demonstrated that dose reductions of irinotecan and treatment delays of oxaliplatin, respectively, were independent negative prognostic factors for progression-free survival. Another study, on platinum-based chemotherapy in patients with ovarian cancer [20], showed that treatment delays, but not dose reductions, were independently associated with worse survival. Furthermore, a randomized phase II trial [21] in metastatic gastric and gastroesophageal junction adenocarcinoma, with standard DCF (docetaxel, cisplatin and uorouracil) given every three weeks compared to reduced dose DCF given every two weeks, demonstrated a six months longer OS in favor of the latter regimen, further emphasizing the importance of having short intervals between treatments. Even though these studies were made on different chemotherapy regimens and in different malignancies, a possible interpretation could be that delayed dosing of platinum-based chemotherapy has a negative impact on outcome, which is also in accordance with our results. However, additional studies are needed to con rm this hypothesis.
The previous literature on RDI in gastric cancer is very sparse. In the metastatic setting Kitagawa et al. [5] found no differences in survival whether RDI of cisplatin and the uoropyrimidine analogue S-1 was above or below 80%. Another study [6] showed that low RDI (89.5% or less) of adjuvant S-1 was an independent predictor of poor disease-free survival. To the best of our knowledge, the present study is the rst to examine RDI, as well as DI and TI, of neoadjuvant chemotherapy in gastric and esophageal adenocarcinoma.
A major limitation of this exploratory, retrospective study is the relatively small sample size with few events, precluding multivariable analyses. Thus, our results should be interpreted with caution and considered mainly as hypothesis generating with a need for validation in additional studies. Moreover, we did not assess the impact of DI and TI for the individual drugs (although TI is usually the same) and it is possible that larger studies could identify cutoffs for the separate chemotherapy components. Based on a preliminary report in 2017 on the now published FLOT4-AIO trial [22], EOX has been gradually replaced by FLOT ( uorouracil, oxaliplatin and docetaxel) as the new standard perioperative regimen, and it would be of interest to assess DI and TI in patients receiving FLOT, which shares the common backbone of oxaliplatin and a uoropyrimidine with the EOX regimen.
Given that side-effects of chemotherapy commonly result in dose reductions and/or treatment delays, the paucity of reports in the literature about how this affects outcome is somewhat surprising. Whether the dose should be reduced or the treatment delayed, in case of non-manageable toxicity, is a clinically very common and important question that needs to be further addressed, not only in this speci c disease setting, but also in other malignancies and treatment situations.

Conclusions
Our results suggest that treatment delays should be avoided in patients receiving neoadjuvant chemotherapy for gastric and esophageal adenocarcinoma. whereby the committee waived the need for consent other than by the option to opt out.

Consent for publication
Not applicable.

Availability of data and material
All data generated or analyzed during this study are included in this published article [and its supplementary information les].

Competing interests
The authors declare that they have no competing interests.

Funding
This study was supported by grants from the Swedish Cancer Society, the Swedish Government Grant for Clinical Research (ALF), the Mrs Berta Kamprad Foundation, the Swedish Society for Gastrointestinal Oncology (GOF), Lund University Faculty of Medicine and Skåne University Hospital Funds and Donations. The funding sources had no role in study design, data collection, data analysis, data interpretation or writing of the manuscript.
Authors' contributions DB: conception and design, acquisition of clinical and treatment data, statistical analyses and manuscript drafting. CH: histopathological re-evaluation, assessment of histopathologic response, manuscript drafting. KJ: conception and design, manuscript drafting. AJ: conception and design, manuscript drafting. All authors read and approved the nal manuscript. Tables.pdf TableS1.pdf