ICI could provide durable antitumor activity in advanced stage NSCLC patients [21,22,23,24]. It was reasonable to stipulate that ICI could have similar durable antitumor activity in operable NSCLC as well. The unique feature of pCR patients was ideal to test long-term antitumor activity of ICI in operable NSCLC. Our pilot retrospective study preliminarily compared tumor recurrence between neoadjuvant ICI/CT and neoadjuvant CT alone in the pCR patients. We found that in NSCLC patients with pCR, neoadjuvant ICI reduced tumor recurrence rate after surgery. Our finding suggested that like in advanced stage NSCLC, neoadjuvant ICI may exert similar long-term antitumor activity in operable NSCLC.

The principal finding of the current study was that in NSCLC patients with pCR, neoadjuvant ICI reduced tumor recurrence rate after surgery. To the best of our knowledge, there was no previous report. This reduced recurrence might be caused by the long-term antitumor activity of ICI, which was mainly observed in advanced NSCLC. Before the advent of immunotherapy, patients with advanced lung cancer had poor prognosis, with a 5-year OS rate of only 6.9% in America between 2010 and 2016 [22]. After the advent of immunotherapy, a growing body of evidence showed that ICI with/without CT brought long-term antitumor activity in some patients with advanced NSCLC [22,23,24,25]. For example, results from the phase III KEYNOTE-024 study suggested that the 5-year OS rate in the pembrolizumab group (31.9%) was approximately double of that in the CT group (16.3%) in previously untreated NSCLC patients with high PD-L1 expression [22]. Similarly in previously untreated, metastatic squamous NSCLC, the 5-year OS rate of the pembrolizumab plus CT group was much higher than that of CT group (18.4% versus 9.7%) [23]. In pretreated patients with advanced NSCLC, nivolumab treatment resulted in a 5-year OS rate of 16% [24]. And the long-term durable anti-tumor effect of ICI might lie in their abilities to induce long-lasting anticancer immune responses, resulting in long-term complete responses [26]. Now evidence altogether suggested that the long-lasting anticancer immune responses of ICI therapies might be related to the generation of anticancer memory T cells. Although the follow-up time of our study was relatively short, our study provided preliminary evidence to show that there might be long-term antitumor activity of ICI in operable lung NSCLC as well. To further verify this finding, prospective studies with longer follow-up time were warranted.

Studies suggested that pCR should not be thought as a complete cure. Recurrences detected during follow-up might be caused by either tumor metastases before or during neoadjuvant therapy or technically undetected living tumor cells. The recurrence rate of pCR patients in the literature varied wildly. Some studies reported alarmingly high recurrence rate of 21.1–46% after surgery [8,9,10]. In pCR patients, recurrence was a major risk factor affecting survival. The 5-year OS rate of pCR patients with recurrence was significantly lower than that of patients without recurrence (19.3% vs. 78.2%) [9]. The time from the end of the neoadjuvant therapy to the surgery and the type of neoadjuvant treatment were the independent risk factors affecting the recurrence [9]. Compared with neoadjuvant CT or RT alone, neoadjuvant RT plus CT had lower recurrence. This was similar to our finding which showed that neoadjuvant ICI plus CT had lower recurrence. This indicated that addition of new neoadjuvant modality to CT could reduce tumor recurrence after surgery.

One major strength of current study was that the pCR patients were our study population. In operable NSCLC patients, the pCR rate after surgery was used to be relatively low, and the neoadjuvant use of ICI significantly increased the rate. Neoadjuvant ICI plus CT had the pCR rate varying from 17.2 to 40.7% among different studies, compared with a pCR rate of 1–5.7% brought by neoadjuvant CT alone [5, 18,19,20]. A meta-analysis showed that pCR rate of neoadjuvant ICI/CT was significantly superior to neoadjuvant CT alone [6]. In the past, the small size of pCR patients made it unfeasible to conduct conclusive study in this specific population. Now the increasing number of pCR patients provided golden opportunity for us to conduct in-depth research and answer some question. By definition, after thorough evaluation of resected specimens, the pCR patients were considered to lack of any viable tumor cells [7]. This tumor cells-free status indicated that all included patients in the current study had same cancer status after surgery, which provided similar subjects to test the long-term effect of ICI. Another strength of current study was that patients who received adjuvant ICI after operation were excluded from current study. The exclusion of those patients eliminated the potential bias brought by adjuvant ICI. So the selection of pCR patients and exclusion of patients with adjuvant ICI increased the credibility of our research results.

Our study revealed that the ICI/CT group had significantly lower percentage of patients who received adjuvant CT than CT group (75% vs. 100%). In pCR patients, it was still unclear if adjuvant therapy after surgery could improve survival. Lococo et al. reported that the pCR patients who underwent adjuvant treatment had better 5-year OS rate, long-term survivals and disease-free survival [8]. In another study of 62 pCR patients, the death risk was estimated to be 3 times higher for patients who did not receive adjuvant therapy [11]. On the contrary, in the review of 759 stage I-III NSCLC patients who achieved pCR after multimodal therapy, the authors did not observe a statistically significant difference in terms of long-term survival between patients with adjuvant treatments and those without [12]. Similarly, Melek et al. and colleagues reported that in a series of 72 pCR NSCLC patients, adjuvant therapy did not influence the long-term outcome [13]. Although the adjuvant treatment in patients with pCR remained controversial in terms of theoretically having no viable tumor cells, the majority of patients in the current received adjuvant CT therapy. In our study, although the ICI/CT group were less likely to receive adjuvant CT than the CT group, the ICI/CT group still had lower tumor recurrence. This disparity lend further support to the long-lasting anti-tumor effect of ICI in operable like in advance stage NSCLC. Moreover, in order to rule out the potential interference of adjuvant ICI, all patients receiving adjuvant ICI were excluded from current study. So whether adjuvant ICI could reduce the recurrence was not within the scope of this article.

Our study also found that the ICI/CT group had significantly lower risk of anemia than the CT group. This may be partly explained by the fact that the ICI/CT group were less likely to received adjuvant CT. The majority of cancer patients would develop anemia during the course of their disease or treatment, and patients receiving CT are especially at risk [27]. By a study of European cancer patients, 75% of patients treated with CT were anemic at least once during the 6-month follow-up, compared to 67% of all patients included [28]. The incidence of anemia in patients receiving their first cancer treatment was 62.7% in patients treated with CT compared to 53.7% in patients overall [28]. Moreover, treatment with platinum, which was essential for CT in lung cancer, was an independent risk factor for patients with CT [27]. So in current study, patients of ICI/CT group might have lower risk of anemia because they were less likely to receive platinum-based adjuvant CT.

Unlike prospective clinical trials, the AEs in the current study were retrospectively evaluated based on medical records, and the underestimation of AEs was unavoidable. This made the comparison with other studies troublesome. Several prospective clinical trials reported the safety profile of neoadjuvant ICI. The results of KEYNOTE-671 revealed that neoadjuvant pembrolizumab plus CT had similar AEs risk to neoadjuvant CT alone (96.7% vs. 95.5%) [5]. The pembrolizumab group had higher risk of alanine aminotransferase level increased (12.9% vs. 7.8%) and similar thrombocytopenia (18.7% vs. 18.5%). The Neotorch study preliminarily reported that the incidence of Grade ≥ 3 AEs was 63.4% in the toripalimab arm and 54.0% in the placebo arm, but the details of AEs profile had not been released [19]. The RATIONALE-315 study also reported that the neoadjuvant tislelizumab plus CT had similar incidence of AEs (99.1% vs. 99.6%) and Grade ≥ 3 AEs (69.5% vs. 69.6%), but the details of AEs profile were unknown [20]. A meta-analysis also found that neoadjuvant ICI plus CT and neoadjuvant double-immunotherapy did not increase the incidence of AEs and Grade ≥ 3 AEs [6]. So far the data of AEs of neoadjuvant ICI was limited, and the release of detailed AEs profile would be helpful.

The current study revealed that squamous cell carcinoma was the predominant histology type of pCR patients. Whether squamous cell carcinoma was associated with pCR was still debatable in the published literature. In a study including 38 NSCLC patients with pCR, 60.5% of patients had squamous cell carcinoma [10]. Similarly, in the study of 72 pCR patients, Melek et al. reported that squamous cell carcinoma accounted for 54.2% [13]. Mouillet et al. [12] also noted squamous cell carcinoma as the sole predictor of pCR [14]. Another study which included 39 pCR patients reported that 69.2% of patients with pCR had squamous cell carcinoma [15]. On the other hand, in the review of 759 stage I-III NSCLC patients with pCR, adenocarcinoma was the predominant histology type [12].

There were a few limitations in this study. First, the minimum follow-up time of 20 months was relatively short for lung cancer patients receiving surgery, and longer follow-up time was needed. Second, the present study was a retrospective study, which came with many inherent limitations including selection bias. The retrospective nature of this study was also prone to biases from missing data and reliance on documentation available for review. Third, the dose modifications of chemotherapy agents were based on the discretion of the attending doctors and therefore, not standardized among the patients. Forth, data were missing regarding PD-L1 expression and tumor mutation burden for the majority of the study patients, so we were unable to evaluate these factors as biomarkers.