Neutrophil to Lymphocyte Ratio (NTLR) Predicts Local Control Failure and Overall Survival after Stereotactic Body Radiotherapy (SBRT) In Metastatic Sarcoma

The neutrophil to lymphocyte ratio (NTLR) and absolute lymphocyte count (ALC) recovery are prognostic across many cancers. We investigated whether NLTR predicts SBRT success or survival in a metastatic sarcoma cohort treated with SBRT from 2014 and 2020 (N = 42). Wilcox Signed Rank Test and Friedman Test compare NTLR changes with local failure vs. local control (N = 138 lesions). Cox analyses identified factors associated with overall survival. If local control was successful, NLTR change was not significant (p = 0.30). However, NLTR significantly changed in patients local failure (p = 0.027). The multivariable Cox model demonstrated higher NLTR before SBRT was associated with worse overall survival (p = 0.002). The optimal NTLR cut point was 5 (Youden index: 0.418). One-year overall survival in SBRT metastatic sarcoma cohort was 47.6% (CI 34.3%-66.1%). Patients with an NTLR above 5 had a one-year overall survival of 37.7% (21.4%-66.3%); patients with an NTLR below 5 had a significantly improved overall survival of 63% (43.3%-91.6%, p = 0.014). Since NTLR at the time of SBRT was significantly associated with local control success and overall survival in metastatic sarcoma treated with SBRT, future efforts to reduce tumor inhibitory microenvironment factors and improved lymphocyte recovery should be investigated.


INTRODUCTION
Sarcomas are a group of rare mesenchymal tumors that account for less than 1% of adult solid malignancies, but represent 21% of pediatric solid malignancies 1 . While the incidence is low relative to other malignancies, their biological heterogeneity and relative treatment resistance have made sarcomas challenging to treat 2 . Sarcomas can be divided into soft tissue sarcomas and sarcomas associated with bone primary tumors (e.g. osteosarcoma and Ewing sarcoma). Advances in sarcoma staging and therapy have led to survival improvements in both classes of sarcomas as well as improved risk-strati cation 3 .
For example, the 5-year survival in rhabdomyosarcoma has improved from around 50-70%. 4 Advancements in chemotherapy have improved prognosis in patients with low grade tumors. However, the success of surgery, radiotherapy, and chemotherapy capable of treating advanced metastatic disease is low 5,6 . Many patients present with high-grade histopathological features, and up to 25% of patients present with distant metastasis 3,4 . The current standard of care for most sarcomas is a combination of chemotherapy and local control that includes surgical resection and/or radiotherapy. Radiation is typically reserved for unresectable disease, when surgical morbidity could limit function, or when metastases are present 7 . There is an unmet need for more data to improve quality of cancer care and survival 8 .
Stereotactic body radiation (SBRT) is a relatively new advancement in radiotherapy that uses image guidance to precisely deliver high doses of ablative radiation in fewer fractions compared to traditional dose regimens. While contemporary indications for SBRT are rapidly expanding, it is most commonly used to treat inoperable early stage non-small cell lung cancer, prostate cancer, unresectable pancreatic cancer, pulmonary and hepatic metastases, and metastases to the spine or brain via a stereotactic radiosurgery approach 9 . Some sarcomas are relatively radio-resistant-that is, resistant to conventionally fractionated radiotherapy without concurrent chemotherapy-and would theoretically bene t from the high biologically equivalent doses (BED) accompanying SBRT regimens. A multi-institutional phase II trial showed a 95% rate of lesion-speci c local control at 6 months and improved progression free survival (PFS) and OS in patients with unresectable metastatic sarcoma treated with SBRT 10 . While local control with SBRT is encouraging, careful consideration of the therapeutic window must be applied as the higher fractional doses of radiation from SBRT increase the risk of developing late complications in normal tissue. A retrospective analysis of 31 patients found similarly high rates of local control and minimal toxicity, with only 1 out of 31 patients experiencing a late grade 3 radiation related toxicity 11 . SBRT has typically been reserved as salvage therapy for treatment resistant disease, so its role in primary treatment remains unknown, but has been investigated on a large cohort in a completed Children's Oncology Group clinical trial in patients with metastatic Ewing sarcoma (NCT02306161).
Another recent advancement in oncology is our understanding about the role of the immune system in cancer. Malignant cancer cells can produce neoantigens that are recognized by the immune system as "not self." Cytotoxic T-cells can identify tumor cells through these foreign antigens and initiate immunemediated tumor cell death 12 . Interestingly, it has been shown that tumors with higher mutational burden can express tumor-associated neoantigens that lead to a better response to immunotherapy 13,14 . In June 2020, the Food and Drug Administration (FDA) approved the use of immune checkpoint inhibitors (ICIs) in solid tumors of any type with a tumor mutational burden of 10 or more mutations per megabase. A myriad of factors including but not limited to cellular immunity and tumor inhibitory microenvironment affect the immune system's ability to eradicate cancer 15 .
Radiation therapy induces DNA damage through direct interaction and generation of reactive oxygen species. Tumor cells that survive the direct toxicity can express new antigens that may be recognized by the host immune system 16 . After radiation induced cell death of cancer cells, neutrophils and lymphocytes are actively recruited to the tumor site 17 . Given the interactions between radiation and the immune system, there may be a promising role for immunotherapy and radiotherapy synergism in the treatment of malignancy. SBRT, rather than conventional radiotherapy, may especially be synergistic with the patient's immune system because it causes less lymphocyte depletion 18 . In pediatric osteosarcoma and Ewing sarcoma, improved absolute lymphocyte count (from either lymphocyte resilience or earlier recovery of lymphocytes post chemotherapy) has demonstrated survival bene ts [19][20][21][22] . If the immune system is a critical part of the radiation e cacy, it would be helpful to have biometrics that help characterize the state of one's immune system prior to SBRT.
Currently, there is no gold standard biomarker or biometric that can be used to assess how a patient's immune system may respond to radiation. The NTLR is a promising candidate that has been correlated with survival in many malignancies including sarcoma before surgery 23 and after chemotherapy 24 .
Elevated pretreatment NTLR have been associated with poorer prognosis in both solid and hematological malignancies [25][26][27][28] . In soft tissue sarcoma, it has been associated with poorer survival and increased risk of distant metastasis 29 . Lymphopenia is a prognostic factor in advanced sarcomas, too 30 .
How the NTLR biometric might change with SBRT has not been investigated. While the literature suggests that NTLR will likely be prognostic across cancer diagnoses, it has not been studied in this population of metastatic sarcoma patients treated with SBRT. In this study, we show how NTLR changes over the SBRT treatment course and optimize a NTLR cutoff score to stratify patients with metastatic sarcoma into high and low risk survival groups. Furthermore, we demonstrate high NTLR was associated with a signi cantly worse prognosis and local control failure after SBRT in patients with metastatic sarcomas.

METHODS
Cohort. This study was approved By the Cleveland Clinic Institutional Review Board (IRB) as a singleinstitution registry of patients with metastatic sarcoma diagnosed between 2014 and 2020 treated with SBRT. After discussion of indications, risks, bene ts, and alternatives, all patients or legal guardians provided written informed consent for SBRT as part of standard clinical care. No additional blood samples were obtained, therefore this was not considered an interventional study requiring additional informed consent for additional research blood samples as per IRB guidelines for a retrospective study. All research was performed in accordance with relevant guidelines and regulations. Included patients had histologically con rmed sarcoma including clear cell sarcoma, desmoplastic small round cell tumor, Ewing sarcoma, myxoid liposarcoma, osteosarcoma, rhabdomyosarcoma, undifferentiated pleomorphic small round blue cell sarcoma, and synovial sarcoma) with at least one metastatic lesion treated with SBRT. CBC with differential data was collected by retrospective chart review at time points prior to SBRT, after SBRT, in follow-up, and at time of progression.
Treatment. Patients with histopathologic diagnosis of sarcoma were treated with SBRT de ned as a fractional dose of ≥ 5 Gy delivered in ve or fewer fractions to site(s) of metastatic disease. The intent of therapy -usually either de nitive for oligo-metastatic disease or palliative for durable local control of metastases-as well as the speci c dose and fractionation regimen were determined at the discretion of the treating physician. Technical details of SBRT technique based on lesion location have been described previously 11,31 .
Laboratory investigations. For each lesion treated with SBRT, complete blood count (CBC) data before and after SBRT treatment and in follow up were recorded. Since CBC is not routinely measured immediately before or after treatment, we included values within 3 months of treatment. We also recorded CBC parameters collected within 3 months after any local failure. We calculated a neutrophil to lymphocyte ratio by dividing absolute neutrophil count by absolute lymphocyte count. Friedman Test was used to assess changes in NTLR through SBRT therapy in failed lesions (comparing before SBRT vs after SBRT vs after local recurrence). Wilcox signed rank was used to compare NTLR in locally controlled lesions before SBRT versus after SBRT. These nonparametric statistical methods were chosen over linear regression methods due to the non-normality of the data and limited time points for comparison of the dependent variables.
Imaging. Patients received routine post-treatment imaging follow-up with CT, MRI, or PET to assess for local control. Local tumor failure was de ned as recurrence of tumor at treatment site as determined by enlarged lesion on CT, MRI, or PET scan and/or interpretation of the radiologist and tumor board members.
Univariate and multivariate analysis of effects on survival. For each patient, the pretreatment CBC parameters within 3 months of SBRT was averaged and included with averaged values with other clinically relevant variables such as concurrent chemotherapy, Karnofsky Performance Status (KPS), and age in univariate Cox regression models to analyze their impact on overall survival. Overall survival was de ned was time to death or last follow up from rst treatment date. A selection criterion of p < 0.15 from the univariate model was used to determine the variables for the multivariate model. This was done because a small number of covariates was more appropriate for the small population size. Age, mean biologic equivalent dose across treatments, KPS, and NTLR were selected for the multivariate model.
A smoothed time-dependent ROC curve calibrated to 24 months of survival was used to generate an optimal cut point for NTLR 32 . Each patient's NTLR was the average of their NTLR values prior to each SBRT treatment session. Optimal cut point to stratify patients into high and low risk groups was determined by the NTLR that maximized the Youden index, de ned as the sum of sensitivity and speci city minus one. The Kaplan Meier method was used to calculate rates of overall survival. We compared the overall survival for these two groups using the log-rank test. All data were stored in a secure RedCap registry, and all analyses were performed in R (version 3.6.3) 33 .

RESULTS
In total, there were 42 patients with 138 lesions that met criteria for the study as detailed in Table 1. The median age at diagnosis was 21 years with a range of 4 to 47 years. The median pretreatment NTLR for the cohort was 5.3. Twenty-ve out of 42 patients had con rmed death, and the median clinical follow up time for surviving patients was 24 months and median imaging follow up time was 7.7 months.   1 Pretreatment complete blood count (CBC) data was charted within 3 months of treating the lesion. 2 Concurrent therapy includes any chemotherapy or immunotherapy received at time of treatment. 3 Local recurrence was determined by CT or PET imaging surveillance. Data is presented as either median [min to max] or as number in category (percent of whole group).   Figure 2 demonstrates the time-dependent ROC curve calibrated to 24 months drawn to assess the optimal cutoff for stratifying high vs low risk groups based on NTLR. The optimal pre-SBRT NTLR cut point value was determined to be 5.033, which was rounded down to 5 and used for strati cation into high risk and low risk groups. 24 patients (57%) were in the low risk NTLR category (< 5). Kaplan-Meier estimates for overall survival were calculated as shown in Fig. 3

DISCUSSION
Our study demonstrated for the rst time a signi cant correlation between increased NTLR and poor local control and signi cantly worse overall survival in high-risk metastatic sarcoma patients undergoing SBRT.
These ndings are consistent with the prior observations in non-small cell lung cancer 34 . Interestingly, patients with lesions with more durable local control after SBRT correlated with a higher baseline NTLR compared to those with lesions that had local failure. However, NTLR in those with lesions that were locally controlled remained stable after therapy and survival seemed to plateau. Among patients experiencing local failure, the NTLR increased slightly after SBRT and then increased signi cantly after con rmed sarcoma recurrence. These ndings suggest that a rising NTLR may be a sensitive biometric of impending recurrence than absolute elevation of baseline NTLR.
However, absolute NTLR was associated with a signi cantly worse survival outcome in patients with metastatic sarcoma treated with SBRT. High pretreatment NTLR was associated with poor survival in the multivariable Cox proportional hazards model controlled for age, dose, and KPS. We also calculated a cutoff NTLR value of 5 to risk-stratify patients with metastatic sarcoma into high risk and low risk groups. An elevated NTLR could re ect either in ammation associated with advancing disease, impaired lymphocyte numbers or resilience (i.e. associated with a low ANC); either these immune parameters would be expected to affect the tumor inhibitory microenvironment and be associated with local control failure 15 .
In oropharyngeal cancer, an association between loco-regional recurrence and elevated NTLR has been observed 35 . In the setting of early stage NSCLC patients treated with SBRT, an elevated NTLR predicted for worse survival, but there was no correlation with local control 34 . This study did not assess changes in NTLR over time, but only evaluated pretreatment variables collected within 2 months of SBRT. Our analysis did not yield a statistical association between time to local recurrence and NTLR prior to treatment in a Kaplan-Meier model; however, our ndings may be limited by the sample size of our study.
NTLR may become a useful biometric to study during immunotherapy with radiation, especially SBRT.
The potential synergy between radiotherapy and immunotherapy is a growing area of research interest 15 . While radiation can result in temporary immunosuppression, it can assist the immune system long term by reducing tumor burden via direct cytotoxicity and exposing neoantigens to the immune system. The induction of neoantigens by radiotherapy and subsequent immunization against those neoantigens have been demonstrated in prostate cancer 36 . SBRT in particular has been demonstrated to stimulate the cellular immune response through enhanced T cell activity, increased antigen presentation, and release of in ammatory cytokines 37 . SBRT mediated damage of one tumor site can lead to systemic activation of the immune system against neoantigens present in other tumor sites leading to an abscopal effect, which has been demonstrated in some clinical settings 37 .
An elevated neutrophil count can be observed in chronic in ammatory conditions, including cancer. Tumor production of cytokines such as IL-1 and IL-6 can induce neutrophilia, with higher circulating neutrophil counts at baseline theoretically corresponding with an increased burden of disease 38 . It has also been shown that increased absolute neutrophil count (ANC) has the potential to suppress cytotoxic T lymphocyte function 39 . The relative lymphopenia associated with high NTLR may be associated with a poorer host immune response against tumor antigens and thus a poorer prognosis. T cells are necessary for checkpoint inhibition therapies and the anti-cancer cellular therapies using CAR-T and CAR-NK 40 . Taken together, an elevated NTLR suggests an increased burden of disease and an impaired immune response, both of which are poor prognostic factors. In the context of cancer therapy, we believe NTLR is a relevant biomarker for assessing potential immune responsiveness following radiotherapy and a useful metric for risk-strati cation.
Our study has several limitations. Since this study is retrospective, it is di cult to ascertain time course of exactly when SBRT alters the NTLR biometric. Given the disease rarity and relatively recent implementation of SBRT for treatment of metastatic sarcoma, the small sample size of the cohort still represents a large series of SBRT treated lesions 11 . Additionally, most sarcoma patients undergoing SBRT were heavily pre-treated with systemic therapies, a factor that likely affected the bone marrow and circulating lymphocytes. Although treatments such as granulocyte colony stimulating factor (G-CSF) could confound results, no one in our SBRT cohort received concurrent G-CSF during or immediately after SBRT.
In conclusion, high NTLR at the time of SBRT treatment was prognostic for signi cantly worse overall Final approval of manuscript: All Authors DATA AVAILABILITY Anonymized datasets recorded and analyzed during the current study are available from the corresponding author on reasonable request.

CONFLICTS OF INTEREST STATEMENT
The authors have no con icts to disclose.

Figure 2
Receiver operating curve (ROC) for overall survival. The ROC was calibrated for 24 months of survival, and the optimal cut point that maximized sensitivity and speci city was a pre-SBRT NTLR value of 5.033.

Figure 3
Kaplan-Meier curve comparing overall survival by NTLR ratio.