Neoadjuvant versus definitive radiochemotherapy of locoregionally advanced oesophageal cancer—who benefits?

Oesophageal cancer (OC) is a moderately frequent tumour entity associated with a poor prognosis. In Germany, the incidence amounts to 9.7/100,000 for men and 2/100,000 for women, with 5‑year survival rates of approximately 18% for each [1]. The major histological subtypes are squamous cell carcinoma (SCC) and adenocarcinoma (AC), whereas the incidence of the latter is rising in the Western World and is associated with increasing rates of obesity [1].

In clinical practice, locoregionally advanced OC is regularly treated with neoadjuvant radiochemotherapy followed by surgery (nRCT-S) according to the CROSS protocol if patients are surgically and functionally operable [2‐5]. However, the superiority of such a trimodal approach over definitive radiochemotherapy without surgery (dRCT) has been controversial for many years, including in randomized controlled trials [6‐11]. Recently, two retrospective studies found no significant difference in overall survival [12, 13].

Nevertheless, there is still an urgent need for additional research data to improve evidence in the multimodal therapy of OC. Thus, the aim of our study was to evaluate our own patient collective regarding differences in survival and recurrence in the course of disease between nRCT‑S and dRCT.

Our study retrospectively collected all patients with locoregionally advanced OC who underwent radiotherapy (RT) of the oesophagus between 2010 and 2018 at our radiation oncology department. The treatment concepts recorded included nRCT‑S and dRCT. Patients who received palliative or adjuvant postoperative treatment were excluded. Data were taken from our institutional databases and clinical information systems. In total, 101 patients were included, 33 receiving nRCT‑S and 68 receiving dRCT. Patient selection is demonstrated as a Consolidated Standards of Reporting Trials (CONSORT) diagram in Fig. 1.

After treatment, patients were regularly followed up every 3 months for 2 years and on a 6-monthly basis thereafter. Missing information about the disease course was obtained by contacting the patients’ oncologists, family doctors or other treating hospitals. The study was approved by the ethics committee of the Hannover Medical School (Nr. 9190_BO_K_2020), and all studies on patient data were carried out according to the principles of the Declaration of Helsinki. All patients provided consent for anonymous statistical analysis in advance. Every patient’s treatment decision was based on a multidisciplinary tumour board’s assessment. Determination of tumour localization and tumour staging included oesophagogastroscopy, endoscopic ultrasound (EUS), biopsy and computed tomography (CT) of the chest and abdomen. All patients were staged in accordance with the Union for International Cancer Control (UICC) classification system 8th edition [14]. The patients’ comorbidities were recorded using the Charlson Comorbidity Index (CCI) [15], whereby weights were assigned to the cancer itself. Nicotine use and alcohol consumption were evaluated as risk factors.

The chemotherapy (CTx) regimens administered were cisplatin + 5-fluorouracil (5-FU; cisplatin 20 mg/m² + 5-FU 1000 mg/m² over 5 days at weeks 1 and 5) as per the Herskovic protocol [16] and later paclitaxel + carboplatin (paclitaxel 50 mg/m² + carboplatin 2 mg × ml⁻¹ × min^–1 area under the curve up to a total of five cycles at weekly intervals). The RT technique used was three-dimensional conformal radiotherapy (3D-CRT), contouring the target volume and organs at risk. Treatment planning of all patients was performed in compliance with the Quantitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC) dose constraints for organs at risk (e.g., V_20Gy < 30% for both lungs and V_30Gy < 46% for the heart) [17]. The tumour and affected locoregional lymph nodes visible on CT were decisive for the gross tumour volume (GTV). Adding margins (radial: 1 cm, longitudinal: 3 cm) to the GTV created the clinical target volume (CTV), followed by another margin of 0.8 cm creating the planning target volume (PTV) to compensate for setup errors. Dose delivery was performed according to the International Commission on Radiation Units and Measurements (ICRU) reports [18, 19]. Patients were examined by oesophagoscopy 4 weeks after completion of dRCT. In the case of macroscopic persistence of the tumour, an additional boost of 8 Gy in two fractions of high-dose-rate brachytherapy was applied. Brachytherapy was performed after CT-based three-dimensional planning to cover 95% of the GTV of the persistent tumour with 95% of the prescribed dose. 192-Iridium was used in an afterloading technique. For brachytherapy, the Bonvoisin–Gérard oesophageal applicator set and the microSelectron® Digital afterloading platform were used (both Elekta, Stockholm, Sweden).

Since 2012, patients who received nRCT‑S have been treated as per the CROSS protocol with 41.4 Gy and paclitaxel + carboplatin [3]. Previously, those patients mainly received doses of 46 Gy and cisplatin + 5-FU. Regarding the histopathological response to radiochemotherapy (RCT) on the basis of the intraoperatively taken tissue samples, patients who received nRCT‑S were classified into “complete tumour regression” and “incomplete tumour regression” [20]. Patterns of failure in the course of disease were recorded as either in-field or out-field (the latter included distant metastases).

Our study showed significantly longer OS and LRC for patients with oesophageal cancer receiving surgery after neoadjuvant radiochemotherapy than for those receiving definitive radiochemotherapy without surgery. However, the CCI values in the latter group were significantly higher. Nevertheless, in a subgroup of lower CCI values (2–6), the superiority of nRCT‑S regarding OS still trended to be significant. Regarding the differently distributed patient characteristics, few patients with cT2- and cT4-stage disease were treated with nRCT‑S due to the relatively high-grade comorbidities in our patients with cT2-stage disease—as low-risk lesions would otherwise rather have been treated by surgery only without any neoadjuvant therapy [24]—and due to the inoperability of the tumour itself in most of our patients with cT4-stage disease. Proximal cervical tumours were all treated with dRCT because of known operational difficulties [25]. Our study showed no significant differences in OS between the histological subtypes of SCC and AC, probably caused by the small number of cases. In the literature, a longer OS for SCC is regularly described [3].

To show the whole patient population treated with curative intent, we did not set inclusion criteria such as performance status limits or freedom of any metastasis (as we included curatively treated oligometastases), as other retrospective analyses did [12]. This might have led to lower survival rates for dRCT in our study compared to other data, while our nRCT‑S rates are similar or even higher [10‐13]. However, survival rates range widely in the literature, probably depending on the study design. In prospective studies, Stahl et al. and Bedenne et al. [10, 11] showed equal results for both treatment approaches, with median OS rates that were considerably lower for nRCT‑S than in our study (16.4 and 17.7 months in nRCT‑S, 14.9 and 19.3 months in dRCT, respectively). In contrast, with a median OS of 49.4 months for patients receiving nRCT‑S, remarkably high survival rates were shown in the CROSS trial [3], making the applied dose of 41.4 Gy the current standard for nRCT‑S. A similarly high median OS of 45 months for nRCT‑S has been shown in a retrospective study by Vitz et al. [26].

Likewise, for median PFS, wide ranges can be found, from 11 to 15.6 to 37.7 months (each for nRCT‑S, in our study, in Haefner et al. [12] and in the CROSS trial [3], respectively). Regarding toxicity, the rate of occurrence of acute adverse events of grades 3 and 4 is mostly described as 10–20% of patients [3, 11, 12], similar to our results. Among our patients, events of acute toxicity occurred more frequently in dRCT, probably due to the higher radiation doses applied.

In the nRCT‑S approach, no considerable incidence of D30 was seen. This is consistent with the finding of Dähn et al. that RT does not cause an increase in perioperative mortality if modern RT techniques are used and certain constraints are consistently followed in RT planning [27].

In terms of CTx, the dRCT group included patients not receiving any CTx because of high-grade comorbidities, underscoring the differences in general health constitution between the treatment groups. Similarly, compliance with full-dose CTx was lower in our study than in other studies [10, 12, 13] due to the high incidence of comorbidities.

There is an ongoing debate about the ideal treatment for locoregionally advanced OC in the literature. Regarding the RT dose for nRCT‑S, Paireder et al. recently considered a modification of the predominant CROSS regimen up to a dose of 46 Gy without disadvantages in safety and effectiveness [3, 28]. For dRCT, de Vos-Geelen et al. emphasized maintaining the dose to ≤50.4 Gy to ensure less toxicity and comparable effectiveness [29], which was confirmed by the randomized controlled ARTDECO study published in 2021 [30]. Moreover, the continuation of treatment for patients with complete remission (CR) after neoadjuvant RCT remains uncertain. While some publications consider CR to be an indication for omitting the operation and instead performing a dose escalation of RCT, others still prefer surgery [6, 31]. This possibly multistep approach raises the question of how response can be measured in routine clinical practice. The current multicentre randomized SANO study compares surgery versus active surveillance for patients with CR [32, 33]. The diagnostic methods used in this study include endoscopy with biopsies in the first evaluation after nRCT‑S and 18F-fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG-PET-CT) followed by endoscopy with biopsies in the second evaluation [32]. A retrospective study in 2018 proposed 18F-FDG-PET-CT as a sufficient method for response evaluation in routine clinical practice, as it combines metabolic and morphologic evaluations [34]. In contrast, a meta-analysis pointed out that noninvasive imaging, such as CT, PET-CT and EUS, is not sufficiently precise to identify CR [35]. In our department, thus far, the histopathological examination of the intraoperatively taken tissue samples was the only evaluation of response. Within this study, a significantly longer median OS in patients with complete tumour regression was not reached due to the small number of patients in this subgroup. However, research is ongoing in this field, and the results of current prospective studies, such as the SANO study or the PRIDE study, could yield additional findings [32, 36]. Another discussed therapeutic option is salvage surgery, which is considered a valuable approach for patients with persistent or recurrent disease after dRCT or in a stage of active surveillance, as mentioned above. In a large multicentre study published in 2015, salvage surgery after dRCT led to similar OS when compared to nRCT‑S [37].

In summary, it is of utmost importance to choose patients carefully for either nRCT‑S or dRCT. The assessment should include at least age and comorbidities, tumour characteristics and, ideally, clinical response to RCT. For elderly patients it is advisable to consult a geriatric physician within the multidisciplinary tumour board [38]. The contribution of our study to the discussion regarding the optimal treatment for locoregionally advanced OC emphasizes that patients with none or few comorbidities seem to profit from nRCT‑S. However, the trend in recent literature has been towards equal effectiveness of dRCT [10‐13].

Our study showed significantly longer overall survival and locoregional control for patients treated with neoadjuvant RCT followed by surgery than for those receiving definitive RCT, especially for subgroups of patients with few comorbidities; however, all of these studies were subject to the aforementioned limitations. In contrast, recent literature has trended towards the equal effectiveness of both approaches. Further research is needed to clarify the standard of care as well as to establish technical developments and multistep approaches, including active surveillance.