Lymphocyte dynamics during and after chemo-radiation correlate to dose and outcome in Stage III NSCLC patients undergoing maintenance immunotherapy

Abstract

Purpose:

We investigated the dynamics of lymphocyte depletion and recovery during and after definitive concurrent chemoradiotherapy (CCRT), dose to which structures is correlated to them, and how they affect the prognosis of stage III non-small cell lung cancer (NSCLC) patients undergoing maintenance immunotherapy.

Methods and materials:

In this retrospective study, absolute lymphocyte counts (ALC) of 66 patients were obtained before, during, and after CCRT. Persistent lymphopenia was defined as ALC <500/μL at 3 months after CCRT. The impact of regional dose on lymphocyte depletion and recovery was investigated using voxel-based analysis (VBA).

Results:

Most patients (n=65) experienced lymphopenia during CCRT: 39 patients (59.0%) had grade (G) 3+ lymphopenia. Fifty-nine patients (89.3%) recovered from treatment-related lymphopenia at 3 months after CCRT, whereas 7 (10.6%) showed persistent lymphopenia. Patient characteristics associated with persistent lymphopenia were older age and ALC before and during treatment. In multivariable Cox regression analysis, recovery from lymphopenia was identified as a significant prognostic factor for Progression Free Survival (HR 0.35, 95% CI 0.13–0.93, p=0.034) and Overall Survival (HR 0.24, 95% CI 0.08–0.68, p=0.007). Voxel-based analysis showed strong correlation of dose to the upper mediastinum with lymphopenia at the end of CCRT, but not at 3 months after CCRT.

Conclusion:

Recovery from lymphopenia is strongly correlated to improved survival of patients undergoing CCRT and adjuvant immunotherapy, and is correlated to lymphocyte counts pre- and post-CCRT. VBA reveals high correlation of dose to large vessels to lymphopenia at the end of CCRT. Therefore, efforts should be made not only for preventing lymphocyte depletion during CCRT but also for helping lymphocyte recovery after CCRT.

Introduction

Immunotherapy has emerged as a powerful therapeutic option across many tumor types and is currently used both alone and in combination with radiotherapy and/or chemotherapy (1, 2). Immune checkpoint inhibitors (ICI) modulate T cell function and have the potential to augment the host immune response against malignant cells (3–5). Maintenance therapy with ICI after definitive concurrent chemoradiotherapy (CCRT) has become the new standard in the treatment of stage III NSCLC after showing clear benefit in randomized studies (6–8).

In patients with advanced NSCLC, PD-L1 expression has been used as the first potential biomarker for predicting treatment response and outcome (9, 10). In addition to PD-L1 expression, ALC and neutrophil-lymphocyte ratio (NLR) can also provide predictive values for patient outcomes (11,12). It has recently been reported that radiation-induced lymphopenia is an important indicator for determining treatment response and survival in patients with metastatic NSCLC undergoing immunotherapy (13).

Radiotherapy (RT) is a double-edged sword when combined with ICI as it leads to both immune stimulatory and suppressive effects. RT stimulates the immune system by releasing inflammatory cytokines and inducing immunologic cell death (14, 15). On the other hand, lymphocytes are one of the most radiosensitive cell lines (16) and it has been shown that radiation can reduce response to ICI (17). Since curative radiation therapy regimen have a prolonged duration of treatment to high total doses, treatment-related lymphopenia happens even in patients with high lymphocyte counts before RT (18), and it has been correlated to dose-volume parameters and dose to lymphocyte-rich organs (19, 20).

There are several studies of the effect of baseline lymphopenia and NLR on the prognosis of NSCLC patients (21, 22); however, none have investigated the effect of treatment-induced lymphopenia during and after radiotherapy in patients treated with CCRT followed by ICI therapy. The reason radiation-induced lymphopenia is being studied so closely is that, as opposed to baseline lymphopenia and other prognostic factors, there are indications that radiation-induced lymphopenia is strongly connected to the RT dose distribution and dose to specific structures (23–26), and could possibly be mitigated (27–29).

Therefore, the purpose of this study is to investigate the pattern of lymphocyte depletion during CCRT and to evaluate how the recovery of peripheral lymphocytes affects the prognosis of patients with stage III NSCLC receiving CCRT followed by maintenance ICI therapy. Furthermore, we investigate the confounding factors that affect lymphocyte recovery using statistical analysis and correlate the dose distribution to lymphocyte decline and recovery using voxel-based analyses (VBA).

Methods and Materials

Patient selection and treatment protocols

This retrospective study included all consecutive patients treated in the observation period at a single center and was approved by the institutional review board (Protocol number: 2021–4-2021–0019) at Yonsei Cancer Center, Seoul, South Korea. Between November 2017 and February 2020, patients fulfilling the following criteria were included: (1) ≥19 years of age, (2) pathologically proven NSCLC, (3) complete blood counts at baseline and follow-ups performed at our institution, facilitating access via electronic medical records, and (4) having undergone the complete course of CCRT and initiated one year of maintenance immunotherapy. Patients whose ALC data were unavailable were excluded. Patients received definitive RT with cytotoxic chemotherapy; all patients received intensity-modulated radiotherapy (IMRT). The PD-L1 expression status was assessed in all patients. The response evaluation on chest CT was performed within 2–4 weeks after CCRT, and patients who showed no progression of disease were candidates for maintenance immunotherapy for one year. Durvalumab or pembrolizumab were used for maintenance immunotherapy. In patients who showed progression of disease during maintenance immunotherapy, the regimen of systemic treatment was changed; however, they were still included in this analysis.

Data collection

Routine blood tests with ALC were performed before treatment, weekly during CCRT, and also at the end of CCRT. After CCRT, ALC counts up to 3 months following the last day of RT were obtained. Lymphocyte counts of 800–1,000/μL, 800–500/μL, 500–200/μL, and<200/μL were defined as grade G1, G2, G3, and G4 lymphopenia, respectively. In this study, we define recovery from lymphopenia and persistent lymphopenia as lymphocyte count of ≥ and < 500/μL, respectively at 3 months after CCRT, as previously used in similar studies (18, 30).

Additional data obtained included age, sex, clinical staging at the time of diagnosis, histology, and tumor PD-L1 expression. The expression of PD-L1 in lung cancer tissue was determined with the DAKO FDA-approved PD-L1, 22C3 PharmDx protocol using the Dako Automated Link 48 platform. PD-L1 expression was considered positive if > 50% of the viable tumor cells exhibited membrane staining of any intensity.

Outcome evaluation and statistical analysis

Progression-free survival (PFS) and overall survival (OS) were measured from initiation of CCRT to last follow-up or the date of tumor progression and to last follow-up or death respectively. The cumulative survival was calculated according to the Kaplan-Meier method and compared using log-rank tests. Both univariable and multivariable Cox proportional hazard regression analyses were performed to determine the prognostic factors for PFS and OS. A multivariable analysis was performed including the variables with p<0.05 in the univariable analysis. Multiple regression was used to determine the factors associated with ALC 3 months after CCRT. All data were analyzed using the R software package, version 3.5, and the survival package, version 3.2–12, missing covariate data was ignored for analysis.

Voxel-based analysis to correlation dose distributions to lymphocyte depletion

Voxel-based analyses were performed based on the methodology developed by Palma et al (31, 32). First, CT images of all patients were deformably registered to a common coordinate system and the corresponding transformation matrices were used to register all dose distributions to a common reference frame. Then, EQD2 maps were adjusted voxel-by-voxel using a generalized linear model (GLM) correcting for all the variables that showed a correlation with lymphocyte depletion with p-value <0.1. The Wilcoxon Rank-sum test was used to test for differences in dose of patients that had post-CCRT ALC <500/μL for each voxel, excluding voxels lying within the tumor volume. The resulting statistic map was enhanced by threshold-free cluster enhancement (TFCE) reflecting neighborhood connectivity of the signal.

Keeping this original result aside, a permutation test was applied on the patient level to correct for multiple comparisons. For each of 10³ permutations, patient labels were randomly shuffled and the same Wilcoxon Rank-sum test and TFCE were applied. The maximum statistic across all voxels was picked for each permutation, and the significance of the original data was interpreted as a quantile function of this distribution of maximum statistics. The schematic workflow describing our analysis is shown in Figure 1.

Fig. 1.

Schematic workflow of the voxel-based analysis. Spatially normalized dose distributions are statistically analyzed by permutation test in order to get the voxel-wise p values and the resulting signal is enhanced by TFCE.

Results

In total, 66 patients with stage III NSCLC received definitive CCRT and maintenance immunotherapy and were included in this analysis, see Table 1 for a description of the cohort and treatment characteristics. At the time of diagnosis, the median ALC was 1,930/μL (range, 540–5,840/μL). Only three patients showed an ALC count<1,000/μL before treatment.

Table 1.

Baseline and treatment characteristics of the patient cohort

Baseline characteristics		Total N=66	(%)
Age	Median (range)	65	(31–85)
Sex	M	54	82
	F	12	18
Smoking history	Never	28	42
	Ex-smoker	26	39
	Current	12	18
T stage	1	9	14
	2	13	20
	3	20	30
	4	24	36
N stage	0	7	11
	1	8	12
	2	24	36
	3	27	41
AJCC 8ht	IIIA	20	30
	IIIB	30	46
	IIIC	13	20
Pathology	Adenocarcinoma	31	47
	Squamous cell carcinoma	34	52
	Others	1	2
PD-L1 expression	0	1	23
	1~25	15	30
	26~50	20	6
	51~75	4	8
	76~100	5	33
Pre-RT ALC (/μl)	Median (range)	1,930	(540–5,840)
	< 2,000	39	59
	≥ 2,000	27	41
Treatment characteristics (N=66)		Total N=66	(%)
RT dose (cGy)	< 6,000	3	5
	6,000	32	48
	6,050	1	2
	6,300	17	26
	> 6,300	13	19
RT days	Median (range)	42	(37–61)
CCRT regimen	Taxol/Carboplatin (TC)	44	67
	IO during CCRT	22	33
	Alimta/carboplatin+durvalumab	7	11
	TC+durvalumab	11	17
	Etoposide/Cispltin+durvalumab	4	6
Maintenance IO	Durvalumab	62	94
	Pembrolizumab	4	6

RT regimens of 60 Gy or 63 Gy in 30 fractions were most commonly used (73.3%); three patients (4.5%) received less than 60 Gy, and the median total dose was 60 Gy. Fraction sizes were 180 to 220 cGy (median 200 cGy). The majority of patients received CCRT with a weekly taxol/carboplatin regimen (n=44, 66.7%), while the others received immunotherapy with cytotoxic chemotherapy during CCRT. The most commonly used maintenance immunotherapy agent was durvalumab (93.9%).

CCRT significantly affected the lymphocyte count as shown in Figure 2A. Most patients (n=65) experienced lymphopenia during CCRT; 32 patients (48.5%) developed G3, and 7 patients (10.7%) G4 lymphopenia. However, 59 patients (89.3%) recovered from treatment-related lymphopenia within 3 months after CCRT. Factors associated with ALC at 3 months after CCRT were a pre-treatment ALC (p=0.034), and ALC at the last week of CCRT (p=0.029) (Table 2).

Fig. 2.

A) The lymphocyte count before, during and after concurrent chemoradiotherapy (CCRT). B) Voxel-based analysis of the differences between patients with and without G3+ lymphopenia 3 months post-RT.

Table 2.

Covariates associated to ALC at 3 months by multiple regression.

Variables		Std. Error	t value	p
Age		0.005	−1.7	0.09
T stage	≥ T3	0.126	−0.3	0.78
N stage	≥ N3	0.136	1.4	0.16
Pre-RT ALC		0.084	2.1	0.03
Last week of RT ALC		0.262	2.2	0.03

After median follow-up of 25.4 months, median PFS was 20.3 months and median OS was not reached; 1-year OS was 86.3%. In all, 34 patients showed disease progression, of which 22 patients (64.7%) developed disease progression during maintenance immunotherapy (within 1 year after CCRT).

In the univariable Cox regression analysis, older age (HR 1.06, 95%CI 1.0–1.1, p=0.038) and recovery from lymphopenia 3 months after CCRT (HR 0.21, 95%CI 0.08–0.60, p=0.003) were significantly associated with OS, and recovery (HR 0.36, 95%CI 0.14–0.94, p=0.036) with PFS (Table 3). Also, pretreatment Neutrophil-Lymphocyte ratio (NLR) was associated with PFS (HR 1.35, 95% CI 1.02–1.75, p = 0.023) and OS (HR 1.58, 95%CI 1.18–2.12, p=0.002). The persistent lymphopenia group showed significantly worse PFS, 6.1 vs. 23.1 months, and OS, 13.9 months vs. not-reached, compared to the recovered group (Figure 3). The multivariable analysis, including all factors associated with ALC and outcome, revealed that recovery from lymphopenia after CCRT was the only significant prognostic factor for OS (HR 0.24, 95% CI 0.08–0.68, p=0.007)and PFS (HR 0.35, 95% CI 0.13–0.93, p=0.034) (Table 3).

Table 3.

Cox survival analysis

		OS
		HR	Univariable 95% CI	p	HR	Multivariable 95% CI		p
Age	≥ 70	1.06	1.00 – 1.11	0.038	1.05	0.99	1.10	0.106
Sex	F/M	2.59	0.59 – 11.25	0.205
T	≥ T3	1.77	0.64 – 4.91	0.274
N	≥ N3	0.67	0.26 – 1.78	0.424
Pathology	Adeno vs. others	1.74	0.68 – 4.43	0.245
Use of IO during RT	Y/N	0.77	0.29 – 2.03	0.594
PD-L1 expression	≥ 50	0.78	0.31 – 1.94	0.591
G3+ during RT	Y/N	0.80	0.32 – 2.00	0.637
Pretreatment ALC	< 2,000	1.67	0.67 – 4.14	0.273
Recovery	Y/N	0.21	0.08 – 0.60	0.003	0.24	0.08	0.68	0.007
		PFS
		HR	Univariable 95% CI	p	HR	Multivariable 95% CI		p
Age	≥ 70	1.00	0.97 – 0.10	0.849	1.00	0.96	1.03	0.776
Sex	F/M	0.81	0.37 – 1.81	0.618
T	≥ T3	0.86	0.43 – 1.74	0.683
N	≥ N3	1.01	0.51 – 2.01	0.975
Pathology	Adeno vs. others	1.02	0.52 – 2.02	0.953
Use of IO during RT	Y/N	0.78	0.38 – 1.61	0.505
PD-L1 expression	≥ 50	0.53	0.25 – 1.09	0.082
G3+ during RT	Y/N	1.08	0.54 – 2.18	0.822
Pretreatment ALC	< 2,000	0.93	0.46 – 1.90	0.851
Recovery	Y/N	0.36	0.14 – 0.94	0.036	0.35	0.13	0.93	0.034

Fig. 3.

The comparison of progression-free survival and overall survival between recovered vs. persistent lymphopenia.

VBA of the dose differences between patients with and without G3+ lymphopenia during the last week of CCRT showed statistically significant differences in dose to large vessels. Figure 2B shows the -log(p) map for the ALC in the last week of RT, areas with visible color-wash (−log(p) >3) indicating significant correlation (p<0.05). Particularly dose to the upper mediastinum surrounding the top of the aortic arch is significantly correlated to lymphocyte depletion. Analyzing the correlation to lymphocyte recovery however revealed no significant dose differences between patients recovering from lymphopenia compared to those that did not.

Discussion

In this study, we found that almost all patients undergoing CCRT experienced lymphopenia, but most recovered within 3 months after CCRT. However, some patients had persistent lymphopenia with an ALC less than <500/μL even at 3 months after treatment, and progression-free and overall survival of these patients was inferior.

This is the first study to investigate lymphopenia at multiple timepoints during and after CCRT and demonstrates the relationship between recovery from lymphopenia and survival in this patient population. At the 2020 World Conference on Lung Cancer, Friedes et al. (28) presented results showing that severe lymphopenia at ICI initiation is related to a more rapid progression in patients with locally advanced NSCLC receiving definitive CCRT followed by maintenance ICI therapy. This is in accordance with our results, as they measured lymphopenia within 2 weeks of ICI initiation, a timepoint similar to ours. However, we performed longitudinal laboratory follow-up and analyzed the association between recovery from lymphopenia as well as severe lymphopenia; and here our study reveals significant differences to previous studies from the pre-PACIFIC era (31, 32), which investigated the impact of lymphopenia after CCRT alone. These studies showed that lymphocyte counts during and at the end of CCRT showed very strong correlations with outcome, while we did not observe this in our study. This could mean that the patients deriving most benefit from adding adjuvant ICI could be those with severely depleted lymphocyte reservoir at the end of CCRT that subsequently recover, which has significant clinical relevance for the management of lymphopenia and follow-up strategies. The effects of prolonged RT regimen have been shown to extend well into the effector phase of the tumor-specific T-cell response (35), and we need to study the immune recovery period after CCRT in more detail and should focus on interventions to help lymphocytes recover in the weeks following CCRT in addition to preventing lymphopenia during CCRT.

Lymphocytes are known to be affected more by RT than by cytotoxic chemotherapy (18, 36), and correlations of radiation dose to specific structures, such as spleen, have been demonstrated (36). Our analysis revealed that lymphocyte recovery is correlated to the ALC at previous timepoints, especially at the end of RT, so we performed a voxel-based analysis to understand the dosimetric factors associated with lymphocyte depletion. As lymphopenia is not related to a single organ, this approach provides more information than investigating correlations using organ-based dose-volume metrics. The quantitative analysis of this correlation between the dose to certain regions and G3+ lymphopenia showed a strong correlation with the dose delivered to the upper mediastinum in the region of the aortic arch. While it should be noted that this region contains the thymus, which is more and more understood to still maintain some low-level activity even in elderly subjects (37,38), no clear conclusions should be drawn from this. Indeed this location is also home to a large number of lymph nodes and there is evidence that these may constitute a tissue reservoir for long-term maintenance of naïve and resting T cells throughout our life span (39), and that blocking egress from lymph nodes dampens immune infiltration into both irradiated primary as well as non-irradiated secondary tumor (35). And lastly this region of significance also contains the largest vessels in the thorax, which is in line with the expectation that dose to large vessels leads to significant irradiation of circulating lymphocytes. However, it is not clear why no significant correlation was observed in the lower mediastinum and the heart itself. This could be related to the fact that the heart is already being spared during RT, and that few patients had inferiorly located tumors that lead to large doses to cardiac structures. Further validation with larger NSCLC patient cohorts is required, as well as comparison to other thoracic malignancies such as esophageal tumors. We want to emphasize that in the voxel-based analysis, dose to the tumor volume was excluded and only dose to normal tissue was taken into account.

A previous study with a computer-based model analyzed the dose to circulating lymphocytes in patients with glioblastomas (34). The model determined that while a single radiation fraction delivered 0.5 Gy to 5% of the circulating cells, after 30 fractions, 99% of the circulating blood had received a dose above 0.5 Gy. Recently, Sung et al. introduced a mathematical model for tumor-immune interactions using clinically extracted data from patients receiving RT, which can be applied to design RT regimen that minimize the lymphocyte-depleting effects of RT. Their model depicts how a shorter treatment regimen leads to less radiation-induced lymphopenia and can reduce the immune recovery period after RT, which can be applied to our patients (40). Analyzing these treatment-related factors could play an important role in improving treatment outcomes and survival in patients receiving ICI therapy.

Although the correlation between lymphocyte count and tumor microenvironment, such as tumor-infiltrating lymphocytes (TILs), was not evaluated in this study, several previous reports suggest that dynamics observed in peripheral lymphocytes are closely correlated with the TILs in the tumor microenvironment (41, 42). Therefore, circulating lymphocytes possibly reflect immunity in the tumor microenvironment and can serve as a biomarker for patients receiving adjuvant ICI therapy that can be easily investigated in clinical practice.

There are several limitations to this study. First, this study included only a small number of patients, and the treatment regimen was similar among all patients, making it difficult to investigate the RT-specific factors further. However, although this study was conducted retrospectively, only a few confounding factors may affect our results; thus, the importance of lymphopenia and the correlation to the dose distribution can be analyzed more clearly, strengthening our analysis. This study was the first to investigate how the dynamics of lymphocyte depletion and recovery during and after definitive CCRT affect the prognosis in patients undergoing ICI therapy. Further studies investigating this crucial recovery period between the end of RT and the initiation of ICI are required, also to elucidate the relationship between circulating lymphocytes and tumor microenvironment and immunity. In conclusion, although many patients with stage III NSCLC experienced G3+ lymphopenia during CCRT, most of them recovered within 3 months after CCRT. For patients who receive maintenance immunotherapy after CCRT, recovery from lymphopenia is essential for improved patient outcomes. Even though lymphocyte counts at the end of CCRT are not directly correlated to outcome, recovery from lymphopenia is correlated to the counts before and during treatment. Furthermore, we demonstrated significant correlations of dose to large vessels in the upper mediastinum with lymphopenia at the end of CCRT. Therefore, clinical strategies to both preserve the lymphocyte count during CCRT and also to support recovery from lymphopenia after CCRT are needed. Further study of the recovery period and association between circulating lymphocytes and tumor microenvironment in patients with NSCLC are required to improve our understanding of the immune effects of CCRT and help us design regimen that maximize efficacy of adjuvant immune modulators.

Highlights.

• Most patients undergoing CCRT and adjuvant ICI experienced lymphopenia

• Some patients had persistent lymphopenia even 3 months after treatment

• Patients with persistent lymphopenia showed inferior PFS and OS

• Lymphocyte recovery is correlated to the ALC at previous timepoints

• Dose to the aortic arch shows strong associations with G3+ lymphopenia none