Prognosis of severe lymphopenia after postoperative radiotherapy in non-small cell lung cancer: Results of a long-term follow up study

Highlights • First study to investigate lymphopenia in NSCLC with postoperative radiotherapy.• Radiation-induced lymphopenia is common is this population.• Lymphopenia is correlated with radiation fraction number and total mean lung dose.• Lymphopenia is an independent risk factor for PFS and OS, particularly in stage III NSCLC.


Introduction
The role of postoperative radiation (PORT) in resectable nonsmall cell lung cancer (NSCLC) remains controversial, particularly in stage III disease [1,2]. The ANITA trial that indicated that patients with pN2 disease could benefit from PORT but patients with pN1 disease did not [3]. However, 95% of patients with clinical stage III-N2 disease receiving PORT were with distant recurrence while only 5% developed an initial isolated local recurrence [2]. Overall survival (OS) was only 43.7% at 5 years, while the locore-gional failure-free survival and distant metastasis-free survival were 53.4% and 42.5%, respectively [4].
Radiotherapy was traditionally thought to be solely a local therapy, but the consensus in the era of immunotherapy is that radiation has systemic effects on the host's immune system. Lymphopenia is a common treatment-related toxicity in cancer patients undergoing radiotherapy. Lymphocytes are highly radiosensitive with their numbers decreasing significantly after radiation and remaining at a low level even for months after CRT [5,6]. For patients treated with chemoradiation for NSCLC, approximately 43% of patients developed grade 3 or 4 lymphopenia after radiation [7]. The incidence of grade !3 lymphopenia increased to 87% after concurrent chemoradiation [8]. Furthermore, emerging evidence indicated that lymphopenia was correlated with inferior survival in several solid tumors such as hepatocellular carcinoma, glioma, esophageal cancer, nasopharyngeal carcinoma, as well as lung cancer [9][10][11][12][13]. Grade !3 lymphopenia was found to be a neg- ative factor for OS (HR 1.5, p = 0.01) in patients receiving definitive chemoradiation for stage III NSCLC [8]. Therefore, determining the effectors of lymphopenia is crucial for further mitigation strategies to protect the host's immune system, which may translate into survival benefits.
Numerous studies have investigated the factors affecting the degree of lymphopenia. In esophageal cancer, proton therapy was superior compared to photon therapy in mitigating lymphopenia [14]. Older age, lower tumor location, greater tumor length, and larger planning target volume (PTV) exacerbated lymphopenia [15]. The nomogram developed in esophageal cancer indicated age, PTV in interaction with body mass index (BMI), radiation technique, and baseline absolute lymphocyte count (ALC) were factors associated with treatment-related lymphopenia [16]. A study in limited-stage small cell lung cancer revealed that patterns of radiation fractionation also affect lymphopenia [17]. In NSCLC patients receiving definitive radiation, larger gross tumor volumes correlated with lower lymphocytes nadirs after radiotherapy [18].
The role of lymphopenia in patients with PORT has not been well studied. Therefore, we conduct a retrospective study to investigate the relationship of patients' characteristics and radiationrelated parameters, as well as baseline ALC, with the risk of severe radiation-induced lymphopenia (sRIL) during CRT in NSCLC patients treated with PORT. The secondary aim was to assess the predictive value of sRIL for clinical outcomes.

Patients
This is an Institutional Review Board approved cohort study in lung cancer patient treated with radiotherapy. Between 1998 and 2017, the medical records from all consecutive patients who underwent surgery followed by radiation (with or without chemotherapy) for NSCLC were extracted. Inclusion criteria were patients who received radiation after surgery who had availability of complete blood count and follow-up data, Eastern Cooperative Oncology Group performance status !2, and no induction chemotherapy. Potential predictors of lymphopenia were recorded, including gender, age, race, BMI, medical complication, smoking status, tumor histology, differentiation grade, tumor location, tumor size, pathological T-and N-stage, pathological stage, and RT-related parameters, such as PTV, radiation modality, total radiation dose (to PTV), fraction number, and lung/heart radiation parameters. Heart or lung V5 was defined as heart or total lung minus-PTV relative percent volumes receiving 5 Gy. All patients were restaged from stage I to III according to the American Joint Committee on Cancer version 7. Patients who underwent interrupted radiotherapy were excluded.

Treatment approaches
All patients treated with PORT by either photon therapy or proton therapy were included. Three-dimensional conformal radiotherapy (3D-CRT), intensity-modulated radiation therapy (IMRT), and volumetric-modulated arc therapy (VMAT) were categorized as photon therapy, while intensity-modulated proton therapy and passive scattering proton therapy were categorized as proton therapy. Two-dimensional radiation was not used in this study. Patients received 50-64 Gy in 25-33 fractions with 1.8 Gy-2.0 Gy per fractionation. In addition, patients who received hyperfractionated radiation twice daily with 1.2 Gy were also eligible. Platinum-based chemotherapy was given peri radiotherapy.

Absolute lymphocyte count assessment
The values of ALC were collected at pre-radiation, during RT, and after 1-month post RT (if applicable). Due to only 16 patients having grade 4 lymphopenia according to the Common Terminology Criteria for Adverse Events version 5.0, lymphopenia was divided into tertiles for the entire population, with sRIL defined as the lower tertile value as absolute lymphocyte counts (ALC) < 0.37 Â 10 3 /ul to minimize the analysis errors.

Statistical analysis
Categorical variables were summarized by frequencies and percentages and compared between the two groups with Chi-square tests or Fisher's exact tests; continuous variables were summarized using means, standard deviations, medians, and ranges and assessed between groups by two-sample t-tests or Wilcoxon rank-sum tests (Kruskal-Wallis tests/ANOVA if appropriate). Generalized linear regressions were used to evaluate the associations between lymphopenia (sRIL versus non-sRIL) and covariates of interest. Unadjusted survival distributions were estimated by the Kaplan-Meier method, and comparisons were made with the logrank test. Cox proportional hazards regression models were used to evaluate the associations between survival outcomes and covariates of interest. All statistical analyses were performed using SPSS version 23.0 (IBM Corp., NY, USA) with 0.05 as a significance level.

Patients
We analyzed a total of 170 patients treated with PORT for NSCLC. The median age for the whole cohort was 62 years (range, 32-89 years). The majority of patients were male (55.3%), white (79.4%), prior/current smoker (83.5%), had adenocarcinoma histology (60.0%), had N2 disease (65.3%) and were clinical stage III (74.1%). Of the 126 patients with stage III disease, 87.3% of patients had N2 disease, while 88.9% were stage IIIA. The median tumor size was 3.9 cm (range, 0.8-14.5 cm). Lobectomy was performed in 79.4% of patients, and 68.8% of patients had a complete resection (R0). 63.5% of patients were given postoperative chemotherapy (POCT) with a median of 4 cycles (range, 1-8). The clinical characteristics are summarized in Table 1.
As shown in Table 1, photon therapy (90.6%) was the dominant RT technique used. The median PTV dose was 54 Gy (range, 48.6-64 Gy) in the whole group, while it was 52 Gy in the sRIL group and 50.4 Gy in the non-sRIL group. For the whole cohort, 36.5% and 22.2% of patients were given 50 Gy and 60 Gy, respectively, while it was 47.0% and 20.9% in non-sRIL, respectively. In sRIL population, 14.5% of patients were given 50 Gy, 50.4 Gy and 63 Gy, respectively. For the dose per fraction in the sRIL group, 70.6% and 22.4% of patients were treated in 2.0 Gy and 1.8 Gy, respectively, while it was 78.3% and 15.7% in the non-sRIL group, respectively.

Lymphopenia during treatment and associated factors
The median interval from the date of surgery to radiation was 1.8 months (IQR, 1.4-2.4), while it was 1.8 months to chemotherapy (IQR, 1.4-3.1). The median interval from surgery to the first time to collect lymphocytes was 1.6 months (IQR, 1.2-2.3). In addition, the median interval between POCT and PORT was 0.8 months (IQR, 0.0-2.5). After surgery but before PORT, the median of ALC for the whole cohort is 1.52 Â 10 3 /ul with ranging from 0.45 À 4.35 Â 10 3 /ul. In addition, before PORT, a total of 23 (13.5%) patients experienced lymphopenia, while 11 patients with grade 1, 10 patients with grade 2, and 2 patients with grade 3, but none of them had sRIL as defined by ALC < 0.37 Â 10 3 /ul. Among these 23 patients, 7 patients developed sRIL in the period of RT, while 2 experienced grade 4 lymphopenia. In the whole groups, a total of 55 (32.3%) patients experienced sRIL during RT while 115 patients had non-sRIL. The median of ALC for the 55 patients was 0.26 Â 10 3 /ul (range, 0.08-0.36 Â 10 3 /ul). Of the patients with Table 1 Clinical characteristics of patient with severe radiation-induced lymphopenia (sRIL) and non-sRIL.

Characteristics
No stage III disease, 29.9% of patients experienced sRIL. Clinical features were well-balanced between the two groups except for surgical margins with significantly more patients with R1/2 (45.5%) in sRIL compared to non-sRIL (24.3%) (p = 0.01). The dosimetric variables were further investigated between the sRIL and non-sRIL groups. As shown in Table 1, the mean PTV and PTV dose in sRIL were 475.6 cm 3 and 58.4 Gy, which were significantly higher than 350.9 cm 3 and 54.5 Gy in non-sRIL (p = 0.015, and p < 0.001). Moreover, the median radiation fraction numbers were also higher in patients with sRIL (32.2 vs. 28.5, p < 0.001). In addition, the mean RT dose, V5, V10, and V20 of total lung were also significantly higher in patients with sRIL (Supplementary Table 1).

Lymphopenia and survival outcomes
The median follow-up for the cohort was 12.2 years (interquartile range, 4.7-14.6 years). 25.9% (44/170) of patients were alive at last follow up. The median PFS and OS for the whole group were 19.8 months and 38.4 months, respectively ( Supplementary  Fig. 1). For patients with sRIL, the median PFS was 14.9 months, whereas it was 21.7 months in the non-sRIL group (p = 0.008, Fig. 1A). For patients with sRIL, the median OS was 28.4 months, which was significantly worse than the 48.3 months in patients with non-sRIL (p = 0.01, Fig. 1B). The 1-, 3-, and 5-year survival rates in the sRIL group were 69.1%, 42.9%, and 28.0%, respectively, in contrast to 79.3%, 57.6%, and 45.5% in the non-sRIL group, respectively.
Univariable analysis indicated that gender, pathological stage, and lymphopenia were correlated with PFS for the whole cohort (Table 3). Multivariable analysis revealed that pathological stage (HR 1.61, p = 0.019) and sRIL (HR 1.68, p = 0.004) were independent predictors of PFS.

Subgroup analysis in patients with stage III NSCLC
Since PORT given to stage I-II patients is mostly due to close or positive surgical margins which is a competing prognostic risk compared to sRIL, we performed a subgroup analysis only in stage III patients. For patients with sRIL, the median PFS was 11 months vs. 18.4 months in patients with non-sRIL (p = 0.015; Fig. 2A). Similarly, the median OS was worse in the sRIL group (20.4 vs. 46.0 months, p = 0.006; Fig. 2B).

Discussion
In the present study, we investigated the factors associated with lymphopenia in patients receiving PORT. Lymphopenia occurred in all patients with 32.3% of patients having sRIL. sRIL was associated with gender, radiation fractionation, and total lung mean RT dose. Moreover, sRIL was significantly correlated with decreased PFS and OS. On a subgroup analysis of stage III patients, sRIL was also associated with poorer survival. Therefore, sRIL appears to be strongly prognostic for PFS and OS in patients with PORT. These findings indicate that lymphopenia is very common in this population and has a negative impact on outcomes. These results suggest that modifications in radiation treatment parameters may mitigate lymphopenia and improve clinical outcomes.
Radiation-associated lymphopenia has been investigated in several studies. Huang et al. showed that severe lymphopenia was correlated with female, old age, lower baseline total lymphocyte count, and higher brain volume receiving 25 Gy in high-grade glioma with radiation plus temozolomide [10]. Susannah et al. showed that RT field size, dose per fraction, and fraction number were correlated with lymphopenia [19]. Our study is consistent with the view that the extent of radiation exposure, regardless of context, is lymphocyte depleting. Increased fraction numbers were associated with sRIL (HR 1.09, p = 0.005). 45.5% of patients with sRIL had a positive surgical margin compared to 24.3% in patients with non-sRIL (p = 0.01). Patients with positive margins had larger treatment volumes and total doses (Table 1; Supplementary  Table 2), which may have contributed to the sRIL. Prolonged RT duration may contribute to severe lymphopenia. Grade 3 to 4 lymphopenia occurred in 54.8% of patients at a median of the 5th week after RT started [20]. Patients with treatment duration >4 weeks had a 28.9% increase in the risk of grade 3-4 lymphopenia, compared to those with the treatment duration of 4 weeks or less (32.1% vs. 62.1%, p = 0.006). In the present study, the median duration of treatment in patients with sRIL is 6 weeks, while it was only 5.4 weeks in patients with non-sRIL.
Notably, the PTV was also significantly larger in patients with sRIL. Lymphopenia is likely caused by radiation exposure to lymphocytes circulating in the body. Larger PTV leads to a larger volume of organ exposed to radiation. Ellsworth et al. has pointed out that circulating lymphocytes should be considered an organ at risk during RT [19]. In NSCLC patients undergoing definitive RT, larger GTVs were correlated with lower lymphocyte nadirs [18]. Other factors that have been shown to be associated with lymphopenia include RT technique, baseline ALC, radiation of immune organ, as well as dosimetric parameters of lung and heart [14,16,18,[21][22][23]. Proton therapy reduced the risk of severe lymphopenia by 29% compared with photon therapy in esophageal cancer with neoadjuvant CRT (17.6% vs. 40.4%; OR 0.29, p < 0.0001) [24]. However, in the present study, we did not see an advantage using proton therapy, likely due to the small sample size and also patient heterogeneity in tumor location and treatment volumes.
Although we didn't find a correlation of sRIL with heart dose, likely due to the fact that most of the postoperative treatment volumes lie superior to the base of the heart and therefore the relative heart dose was quite low, we did find lung dosimetric parameters to be correlated with severe lymphopenia. Total lung V5, V10, V20, and total lung mean dose were higher in patients with sRIL, with the latter being significantly correlated with sRIL on univariable and multivariable analyses. As showed in Supplementary Table 4, total lung mean dose was significantly correlated with sRIL (r = 0.212, p = 0.009). Therefore, only total lung mean dose was analyzed, considering lung V5, V10, V20 and total lung mean dose affect each other. This is in contrast to the Tang et al. study, which analyzed 711 patients receiving definitive CRT for NSCLC and found that lung V5 was significantly associated with lymphocyte nadirs [18]. The difference may be due to the variation of patients enrolled.
Another interesting consideration is whether a lower lymphocyte count due to surgery further contributed to the radiationassociated lymphopenia. A previous study demonstrated that reduced mean lymphocyte count was correlated with thoracic surgery (p < 0.0001) [25]. In a study investigating the association between postoperative lymphopenia and postoperative pneumonia, the lymphocyte nadir was 1.0 ± 0.5 Â 10 9 /L which occurred on day 1 after the surgery; however, the lymphocytes increased gradually after that. Similarly, lymphocytes decreased to 1.1 ± 0.4 9 Â 10 9 /L 3 days postoperatively but recovered on day 4 in advanced oral cancer treated with surgery. The reduction of lymphocytes caused by surgery is limited and typically recovers fairly quickly afterwards. This is not the same as the lymphopenia caused by radiotherapy. Several studies have indicated that lymphocyte counts decline exponentially during radiation, reaching nadir between 3 and 5 weeks from the start of radiotherapy [20,24]. In Table 2 Univariate and multivariate analyses for association with severe radiation-induced lymphopenia. Abbreviations: BMI, body mass index; COPD, chronic obstructive pulmonary disease; Smoking a , prior/current vs. never; pT/N stage, pathological tumor/node stage; AdjChemo, adjuvant chemotherapy; PTV, planning target volume; RT, radiation; 3D-CRT, three-dimensional conformal radiation therapy; IMRT*, Intensity-modulated radiation therapy (including VMAT, Volumetric modulated arc therapy); RTfxNo., radiation fraction number; V5, organ volume receiving 5 Gy; mDose, mean radiation dose; ALC, absolute lymphocyte count.
addition, the lymphocyte counts are not restored fully for nearly half of the patients even 1-2 months after completing radiation therapy [26]. The difference may be due to different types of lymphocyte damage. The decrease in lymphocytes caused by surgery may be caused by the body's acute stress response, while substantial damage from radiotherapy to lymphocytes contributed to lymphopenia. In our study, 13.5% (23/170) of patients experienced lymphopenia after surgery but before radiation; however, none of them undergone sRIL. Furthermore, only 7 of 23 patients developed sRIL during RT. Therefore, it seems that the contribution of surgery on lymphopenia is limited. Previous studies have shown that the severity of lymphopenia is associated with clinical outcomes. Tang et al showed that higher lymphocyte nadirs were associated with prolonged OS (p = 0.01) and event-free survival (p < 0.001) [18]. Ladbury et al. retrospectively reviewed 117 patients with stage III NSCLC treated with definitive CRT and showed that grade ! 3 lymphopenia was correlated with higher estimated dose of radiation to immune cells  Consistent with these studies, we also showed that sRIL is a poor prognostic factor for PFS and OS in stage I-III PORT patients. This study has several limitations as a single-institution retrospective study with a relatively small sample size. We also had to widen the time period for which this data was collected due to the lack of diagnostic lab collections for patients undergoing  PORT alone without chemotherapy. This is especially true in more recent years since only 8.8% (15/170) of patients between 2015 and 2017 had their blood drawn for complete blood count analysis. As a consequence, most of the patient data were collected before 2015, when fewer patients received proton radiotherapy. In addition, the inclusion of patients with stage I-II disease increased the rate of patients with positive surgical margins (31.2%) which confounded the relative impact of sRIL. Defining a separate subset analysis in the stage III disease was a way to better study the impact of sRIL after PORT.
In conclusion, we demonstrated that patients with NSCLC receiving PORT were vulnerable to sRIL. Severe radiation-induced lymphopenia was correlated with total lung mean dose and radiation fractionation numbers and associated with poorer survival outcomes. Future studies will need to address the effects of using more hypofractionated course of radiotherapy or advanced radiation delivery techniques like proton therapy to further reduce the risk of sRIL. These approaches are especially relevant in this era of immunotherapies as these agents are increasingly being incorporated sequentially with radiotherapy.

Funding
This study was funded in part by the Shandong Cancer Hospital for WJ and the National Institutes of Health through MD Anderson's Cancer Center Support Grant CA016672.

Patient Consent Statement
This is a retrospective analysis of previous treatment data on IRB approved protocol RCR05-0967 which waives patient consent of chart review for clinical outcomes analysis in thoracic cancer patients.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.