Clinical Impact of Early Tumour Shrinkage in Metastatic or Unresectable Oesophageal Cancer Treated with Pembrolizumab plus Chemotherapy

Oesophageal cancer ranks 10th among newly diagnosed cancers and 6th in cancer-related mortality worldwide [1]. Patients with oesophageal cancer generally undergo multidisciplinary management, including surgery, radiotherapy, chemotherapy, and immunotherapy [2‒10]. Combination therapy with fluoropyrimidine and platinum-based chemotherapy, most commonly oxaliplatin or cisplatin, has been the mainstay of first-line treatment for metastatic or unresectable locally advanced oesophageal cancer for decades [11‒14], with minimal improvement in overall survival (OS). Therefore, metastatic or unresectable locally advanced oesophageal cancer remains a disease with high mortality, and an effective first-line treatment remains an unmet need for this patient population.

Immune checkpoint inhibitors have shown effective antitumour activity as second-line or later therapies for patients with unresectable, advanced, or metastatic squamous cell carcinoma (SCC) or oesophageal adenocarcinoma (AC) of the oesophagus [15‒17]. More recently, the results of the KEYNOTE-590 trial demonstrated the benefit of immune checkpoint inhibitor therapy combined with chemotherapy for pembrolizumab plus chemotherapy as first-line treatment in previously untreated, histologically or cytologically confirmed, locally advanced, unresectable, or metastatic AC or SCC of the oesophagus, or Siewert type 1 gastroesophageal junction AC [8]. Based on these results, the Japanese Oesophageal Society indicated that there is strong evidence to recommend pembrolizumab plus chemotherapy and cisplatin plus 5-fluorouracil (CF) as first-line therapies for the treatment of unresectable, advanced/recurrent oesophageal cancer [18]. However, the predictive factors for treatment efficacy remain controversial.

Early tumour shrinkage (ETS) has been studied as a predictor of the clinical outcomes of first-line treatment in patients with various types of metastatic cancer [19‒25]. The depth of response (DpR) has been also reported as a continuous measure of efficacy that may serve as a predictor of long-term treatment outcomes [26] and is associated with progression-free survival (PFS), OS, and post-progression survival in patients with metastatic cancer receiving first-line treatment [23, 24, 27]. In oesophageal cancer, Ura et al. [28] previously investigated the association between ETS and clinical outcomes in patients with metastatic oesophageal cancer treated with chemotherapy. More recent studies identified ETS as a highly promising imaging biomarker, specifically for outcomes in patients treated with immunotherapy alone [29‒31]. However, the role of ETS/DpR in combination treatment with chemotherapy plus immunotherapy in patients with metastatic cancer still remains unclear. In addition, no study has investigated ETS and DpR in predicting clinical outcomes in patients receiving immunotherapy regimens for oesophageal cancer. This study investigated the clinical utility of ETS and DpR as a predictor of survival outcomes in patients with metastatic or unresectable oesophageal cancer treated with pembrolizumab plus CF.

This study considered 54 consecutive patients with metastatic or unresectable locally advanced oesophageal cancer who received pembrolizumab plus CF therapy at the Osaka International Cancer Institute between December 2021 and February 2023. The study population included treatment-naïve patients with advanced or metastatic oesophageal cancer and those who had received any prior treatment. All patients had measurable disease according to the Response Criteria in Solid Tumours (RECIST) version 1.1 [32]. Among them, 1 patient was excluded because their performance status worsened before the initial therapeutic evaluation. Therefore, the present study enrolled 53 patients. Their characteristics, clinicopathological features, and imaging findings were obtained from medical reports. The patients were evaluated by oesophagoscopy, computed tomography, and positron emission tomography. The study protocol was approved by the Human Ethics Review Committee of Osaka International Cancer Institute (18033-4, September 24, 2021).

The patients received pembrolizumab (200 mg) plus chemotherapy (5-fluorouracil 800 mg/m² on days 1–5 plus cisplatin 80 mg/m² on day 1 for a maximum of six cycles) intravenously once every 3 or 4 weeks. Treatment was continued until disease progression, unacceptable toxicity, illness, physician or patient decision to withdraw, noncompliance, complete response (CR), or discontinuation for administrative reasons.

Tumour response was routinely assessed per RECIST version 1.1 by the investigators every 2 cycles of treatment. A computed tomography slice thickness of 5 mm was used for tumour evaluation. Disease progression was verified by several investigators. Adverse events were evaluated throughout and after treatment discontinuation and graded by several investigators according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 4.0. ETS was defined as the percentage decrease in the sum of the target lesion sizes at the first evaluation compared to that at baseline. DpR was defined as the percentage of maximal tumour shrinkage in the target lesions during treatment compared with that at baseline. PFS was calculated from the date of diagnosis of metastatic or unresectable disease to the date of disease progression or death, whichever occurred first. OS was calculated from the date of diagnosis of metastatic or unresectable tumours to the time of death due to any cause or the last follow-up.

The results were expressed as medians (ranges) for continuous variables and percentages for categorical variables. Safety was assessed for all patients who received at least one dose of the study treatment. The Kaplan-Meier method was used to estimate the OS and PFS. Between-group differences in OS and PFS were assessed using stratified log-rank tests. The ETS and DpR cut-off values were determined based on hazard ratios (HRs), 95% confidence intervals (CIs), and p values of PFS. Cox proportional-hazards regression models were used for univariate and multivariate analyses of PFS. Since ETS and DpR were expected to be confounding factors, the factor used in model A and B were changed. Statistical analyses were performed using JMP^® Pro 16.0.0 (SAS Institute Inc., Cary, NC, USA).

The baseline characteristics of the patients are summarized in Table 1. The median age was 65 years (range 42–83 years), with most patients being men (41 patients, 77%). Most patients had an Eastern Cooperative Oncology Group performance score (ECOG PS) of 0 (35 patients, 66%) and nearly all had SCC (51 patients, 96%). Twenty-seven patients (51%) were treatment naïve, while 26 (49%) had received any treatment before initiating pembrolizumab plus CF therapy: 10 (19%) had received adjuvant therapy, 8 (15%) chemoradiation, 4 (7%) chemotherapy, and 4 (7%) non-radical surgery. Patients who received prior chemoradiation therapy or chemotherapy developed distant metastasis during prior treatment or were preoperatively diagnosed as unresectable due to tumour growth or invasion into adjacent organs. Patients who underwent non-curative surgery were intraoperatively diagnosed with unresectable tumours due to T4b despite being scheduled for radical surgery after preoperative chemoradiation therapy.

The RECIST evaluation results are summarized in Table 2. The primary response rates were 38% overall, 38% in those with metastatic disease, and 36% in those with unresectable locally advanced disease. The best overall response rates were 53% for all patients, 48% for those with metastatic disease, and 73% for those with unresectable locally advanced disease. A complete clinical response was observed in 13% of all patients, 12% of patients with metastatic disease, and 18% of patients with unresectable locally advanced tumours.

The median follow-up was 10.8 months (range 4.0–20.6 months). The HRs, 95% CIs, and p values of PFS and OS were calculated according to ETS cut-off values of 10%, 20%, 30%, and 40% (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000535186), which revealed an optimal cut-off value of 20% based on after categorization across all outcomes. The Kaplan-Meier curves for PFS and OS between the groups with ETS of ≥20% and <20% are shown in Figure 1a and b. Patients with ETS <20% and ≥20% had median PFS of 12.7 months and 5.5 months, respectively (HR 0.11, 95% CI: 0.05–0.27, p < 0.001). All patients with ETS <20% had disease progression within 1 year of the date of diagnosis of metastatic or unresectable disease. Patients with ETS ≥20% and <20%had median OS durations of 14.4 months and 8.2 months, respectively (HR 0.10, 95% CI: 0.04–0.29, p < 0.001).

The HRs, 95% CI, and p values for PFS and OS were calculated according to DpR cut-off values of 10%, 20%, 30%, and 40% (online suppl. Table 2), which revealed an optimal DpR cut-off value of 30% after categorization of the PFS outcomes. The Kaplan-Meier curves for PFS and OS between groups with DpR ≥30% and <30% are shown in Figure 1c and d. Patients with a DpR of ≥30% and <30% had median PFS of 12.7 months and 4.9 months, respectively (HR 0.10, 95% CI: 0.04–0.23, p < 0.001). All patients with DpR <30% had disease progression within 1 year of the date of diagnosis of metastatic or unresectable disease. Patients with DpR ≥30% and <30% had median OS of 14.4 months and 8.0 months, respectively (HR 0.12, 95% CI: 0.05–0.32, p < 0.001).

Investigation of the size changes in target lesions from baseline according to ETS ≥20% and <20% (Fig. 2) showed that most patients with ETS <20% experienced early tumour growth. Even patients with a partial response (PR) to the primary response by RECIST experienced early regrowth. In contrast, most patients with ETS ≥20% experienced longer tumour responses and retained better partial or CRs.

Figure 3a shows a comparison between ETS and the best overall response. Most patients with ETS <20% experienced progressive disease (n = 9/10, 90%) or stable disease (n = 15/15, 100%) as the best overall response, while those with ETS ≥20% experienced PR (n = 20/21, 95%) or CR (n = 7/7, 100%).

Figure 3b and c show the results of comparisons of ETS according to prior treatment status. In patients with metastatic disease, the median change from baseline in those with non-prior treatment, adjuvant therapy, or prior chemoradiation therapy/chemotherapy was −18.4 (range −100 to 77.9), −18.5 (range −52.7 to 87.7), or −13.4 (range −59.4 to 60.7), respectively, with no significant differences between prior treatment and ETS. The rate of ETS ≥20% was also similar among patients with and without prior treatment (n = 13/26, 50%), adjuvant therapy (n = 3/10, 30%), and prior chemoradiation therapy/chemotherapy (n = 3/6, 50%) (Fig. 3b). Most patients with unresectable, locally advanced disease in the present study had received prior treatment, chemoradiation therapy, or surgery (n = 10/11, 91%). ETS ≥20% was observed in half of patients treated with pembrolizumab plus CF therapy after prior chemoradiation therapy (Fig. 3c).

The potential pre-treatment or therapeutic evaluation factors for PFS were investigated (Table 3). Univariate analysis showed that disease status (unresectable or metastatic), ETS (≥20% or <20%), and DpR (≥30% or <30%) were significant factors for PFS in patients administered pembrolizumab plus CF therapy. Multivariate analysis identified ETS ≥20% as an independent factor for better PFS in patients administered pembrolizumab plus CF therapy (odds ratio [OR] 0.15, 95% CI: 0.06–0.35, p < 0.001). DpR≥30% was also identified as an independent factor for better PFS (OR 0.11, 95% CI: 0.05–0.26, p < 0.001), together with ECOG PS and disease status.

Among patients administered pembrolizumab plus chemotherapy, 18 (34%) experienced immune-mediated adverse events, with interstitial pneumonitis (9%), skin (8%), and hypoadrenocorticism (8%) being the most common (online suppl. Table 3). Among these patients, grade 3 or higher immune-mediated adverse events occurred in 2 (4%) patients, including 1 case each of interstitial pneumonitis and central nervous system dysfunction.

Based on the results of the KEYNOTE-590 trial, pembrolizumab plus chemotherapy has been indicated as the first-line treatment for unresectable or metastatic oesophageal cancer, or Siewert type 1 gastroesophageal junction AC [8]. However, no reports have investigated the predictive factors of clinical outcomes in patients receiving regimens containing immunotherapy for oesophageal cancer. The present study is the first to investigate the clinical utility of ETS and DpR as predictors of survival in patients with metastatic or unresectable oesophageal cancer treated with first-line immunotherapy plus chemotherapy. In this study, ETS and DpR were associated with longer PFS and OS in patients with metastatic or unresectable locally advanced oesophageal cancer receiving pembrolizumab plus CF. Our results demonstrated that an ETS cut-off value of 20% could be a promising on-treatment marker for predicting further sensitivity to pembrolizumab plus CF therapy although only 53 patients were included. In addition, we observed high ETS in patients who received prior treatment, similar to that in treatment-naïve patients.

Only one other study has investigated the association between ETS and clinical outcomes in patients with metastatic oesophageal cancer treated with chemotherapy [29]. Among the 53 patients in the present study, 35 patients with ETS ≥20% (66.0%) had longer PFS and OS than those with ETS <20% (median PFS 7.5 vs. 3.4 months, HR 0.26, 95% CI: 0.14–0.49, p < 0.01, median OS 13.8 vs. 6.1 months, HR 0.20, 95% CI: 0.11–0.39, p < 0.01). In previous studies, although high ETS prolonged the survival among patients treated with chemotherapy alone, they showed insufficient long-term response, with gradually decreasing PFS and OS. In the present study, patients with ETS ≥20% showed a sufficient long-term response (median PFS 12.7 months) compared to the previous study with chemotherapy alone (median PFS 7.5 months). The KEYNOTE-590 trials also showed prolonged duration of response among patients administered pembrolizumab plus chemotherapy (8.3 months vs. 6.0 months) and a high response rate duration of ≥24 months (18% vs. 6%), with a high objective response rate (45.0% vs. 29.3%) [8]. Therefore, a longer response duration can be expected in patients with high ETS treated with pembrolizumab plus CF compared to that in patients with high ETS treated with chemotherapy alone.

ETS can be achieved at a specified early time point after treatment initiation. However, DpR is more likely to be delayed because the time to maximum shrinkage is longer and unpredictable. Therefore, ETS is more useful than DpR as an early on-treatment predictive marker. Among various cancer types, patients with low ETS showed a poorer prognosis than those with high ETS; however, the prognosis of patients with oesophageal cancer with ETS ≥20% was worse than that of patients with other types of cancer [19‒25]. In the present study, all patients with ETS <20% experienced disease progression within 1 year after treatment initiation. In contrast, most patients with ETS ≥20% experienced relatively longer responses and better PR or CRs in some patients. In addition, multivariate analysis showed that ETS ≥20% was an independent prognostic factor for PFS, consistent with a previous report [29]. Therefore, ETS in oesophageal cancer may be a more promising early predictive marker for further treatment effects and survival than in other types of cancer. Even if ETS <20%, early switching to the next treatment depending on the patient’s situation might help avoid adverse events from the first treatment and allow the patient to receive the next treatment with a better PS.

The KEYNOTE-590 trial enrolled treatment-naïve patients [8]; no previous studies have reported on the treatment outcomes of pembrolizumab plus CF therapy in patients with prior treatment. The present study included patients with metastatic or unresectable oesophageal cancer with (49%) and without (51%) prior treatment. In patients with metastatic disease, ETS ≥20% was observed in 50% of patients with prior chemoradiation or chemotherapy and 30% of those with adjuvant therapy, similar rates to those in treatment-naïve patients (50%). Although the prognosis of patients with unresectable tumours is extremely poor [33], >90% of patients with unresectable locally advanced disease receive prior treatment, chemoradiation therapy, or surgery, 70% of whom experienced ETS ≥20%. Moreover, univariate analysis showed that treatment history before pembrolizumab plus CF therapy was not associated with PFS (OR 0.97, 95% CI: 0.50–1.91, p = 0.942). Therefore, patients administered pembrolizumab plus CF therapy might have the potential to achieve a high ETS, even in those with insufficient responses to prior treatments. This is the first study to investigate the clinical outcomes of pembrolizumab plus CF therapy for metastatic or unresectable oesophageal cancer in a real-world patient population, which differs from the target patients of KEYNOTE-590.

This study has several limitations. First, this was a single-centre, retrospective study with inherent selection and confounding biases. Second, the sample size was relatively small. As pembrolizumab plus CF therapy was accepted as a first-line treatment for metastatic or unresectable oesophageal cancer since the end of 2021, reports on the clinical outcomes of pembrolizumab plus CF therapy have been still lacking except for KEYNOTE-590. Third, since programmed cell death ligand 1 (PD-L1) expression, such as tumour proportion score or combined positive score, of patients with metastatic or unresectable oesophageal cancer is not measured routinely in our institute, the association between ETS/DpR and PD-L1 expression could not be evaluated in the present study. Fourth, no established standard cut-off values for ETS for survival outcomes have been previously reported, although values of 10% or 20% are commonly used in metastatic colorectal cancer [19‒24]. The cut-off value might be influenced by the type of cancer, staging, treatment, and efficacy marker of interest to investigators. Thus, further studies are required to determine the optimal ETS cut-off value in patients with metastatic or unresectable locally advanced oesophageal cancer.

In conclusion, ETS and DpR were associated with survival outcomes in patients with metastatic or unresectable locally advanced oesophageal cancer treated with pembrolizumab plus CF. An ETS may be a promising on-treatment marker for predicting further sensitivity to pembrolizumab plus CF therapy and can serve as a valuable therapeutic decision-making tool. Moreover, a longer duration of response could be expected in patients with pembrolizumab plus CF therapy with ETS ≥20% although this study did not include a large number of patients. Our results provide promising information for the early prediction of further treatment efficacy in patients with metastatic or unresectable locally advanced oesophageal cancer. Future studies will be expected to address the limitations of the current study, including multi-centre studies with larger patient populations to validate our findings of the optimal ETS cut-off in this patient population.

We would like to thank Editage (www.editage.com) for the English language editing.

The Human Ethics Review Committee of Osaka International Cancer Institute approved this study (No. 18033-4) and it conforms to the provisions of the Declaration of Helsinki. All patients provided written informed consent for surgery and the use of clinical data as required by the Institutional Review Board of Osaka International Cancer Institute.

The authors declare no conflict of interests for this article.

No funding was received.

Protocol/project development and manuscript writing/editing: T. Sugase and H. Miyata. Data collection or management: T. Sugase, T. Kanemura, T. Takeoka, and N. Matsuura. Data analysis: T. Sugase, T. Kanemura, T. Takeoka, N. Matsuura, Y. Masuike, N. Shinno, H. Hara, T. Omori, M. Kitakaze, M. Kubo, Y. Mukai, T. Sueda, S. Hasegawa, H. Akita, J. Nishimura, H. Wada, M. Yasui, and H. Miyata.

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.