Perioperative neutrophil:lymphocyte ratio and postoperative NSAID use as predictors of survival after lung cancer surgery: a retrospective study

The association between neutrophil:lymphocyte ratio (NLR) and poor long-term outcomes in patients with non–small-cell lung cancer (NSCLC) has been demonstrated in numerous studies. The benefit of perioperative administration of anti-inflammatory drugs on these outcomes has not been well established. Our aim in this retrospective study was to investigate the effects of postoperative nonsteroidal anti-inflammatory drug (NSAID) administration and NLR on tumor recurrence and survival in patients' undergoing surgical resection for NSCLC. This retrospective study included perioperative data from 1139 patients who underwent surgical resection for stages I–III NSCLC. Perioperative data such as baseline characteristics, adjuvant or neoadjuvant therapy, pre- and postoperative NLR, and NSAID use (ketorolac, ibuprofen, celecoxib, or in combination) were included. We evaluated the association between preoperative NLR and NSAID use on recurrence-free (RFS) and overall survival (OS). In all, 563 patients received an NSAID as a part of their postoperative management. The majority of patients received ketorolac (n = 374, 67.16%). Ketorolac administration was marginally associated with better OS (P = 0.05) but not with RFS (P = 0.38). Multivariate analysis (n = 1139) showed that preoperative NLR >5 was associated with a reduction in RFS (hazard ratio [HR] = 1.37; 95% confidence interval [CI] = 1.05–1.78; P = 0.02) and OS (HR = 1.69; 95% CI = 1.27–2.23; P = 0.0003). However, after accounting for tumor stage, NLR ≥5 was a predictor of RFS and OS only in patients with stage I NSCLC. To conclude, preoperative NLR was demonstrated to be an independent predictor of RFS and OS in a subset of patients with early stage NSCLC. Ketorolac administration was not found to be an independent predictor of survival.


Introduction
Lung cancer is still the leading cause of death among women and men in the United States [1]. Surgery remains the mainstay treatment option for patients with non-small-cell lung cancer (NSCLC) [2]; therefore, a number of perioperative-related factors including tumor stage, margin status, type of resection, and type of anesthetic and analgesic technique have been investigated with the goal of predicting and improving long-term survival [3,4].
Neutrophils are key cellular components of the inflammatory system and lymphocytes play an important role in immune surveillance and immune editing. Therefore, the neutrophil:lymphocyte ratio (NLR) correlates those two cellular components as a marker of perioperative inflammation [5,6]. In the context of cancer, an NLR >5 has been suggested as an independent prognostic factor for decreased recurrence-free survival (RFS) in patients with malignancies, which highlights the importance of perioperative inflammation and immune suppression in oncological outcomes [7][8][9][10].
It has been speculated that the use of nonsteroidal anti-inflammatory drugs (NSAIDs) during and after surgery could not only modify the tumor microenvironment in which micrometastasis are present but also reduce migration and invasion of circulating malignant cells; therefore, the administration of these drugs in the perioperative period might have a significant impact on cancer recurrence [11][12][13]. The perioperative administration of NSAIDs (ketorolac or diclofenac) has been shown to be an independent risk factor for distant metastasis-free survival, and ketorolac administration, specifically, was a predictor of better overall survival (OS) in an observational study of patients with stage I or II NSCLC [8].
The aim of this retrospective study was to determine possible associations between preoperative NLR and postoperative use of NSAIDs on RFS and OS in patients with stage I-III NSCLC undergoing curative resection. We hypothesized that patients with a low inflammatory status (NLR <5) and taking NSAIDs postoperatively have a longer RFS and OS.

Methods
Study approval and waiver of written informed consent was obtained from the University of Texas MD Anderson Cancer Centre Institutional Review Board (IRB) prior to the start of the study.
Perioperative data were collected, stored, and managed in a REDCap (Research Electronic Data Capture) database from patients who underwent surgical resection for primary stage I, II, or III NSCLC between January 2004 and December 2010 at The University of Texas MD Anderson Cancer Center Centre was included in this retrospective analysis. Patients 18 years or older who had surgery with the intention to cure were included in the analysis. Those who had palliative surgery or secondary malignancies were excluded from the analysis. The analyzed data included patient age, gender, body mass index (BMI), American Society of Anesthesiology (ASA) physical status, tumor histology, WHO tumor stage (I, II, or III), type of surgery, and administration of neoadjuvant and/or adjuvant chemotherapy and/or radiation. Intraoperative anesthetic care of the patients consisted of general balanced anesthesia typically involving the use of a volatile anesthetic in oxygen, intravenous opioids, muscle relaxation with a nondepolarizing agents, and normothermia.
Postoperative management typically consisted of patient-controlled epidural analgesia with or without the addition of nonsteroidal anti-inflammatory drugs that started within 72 h after surgery. Patients who were administered an NSAID were given ketorolac (30-60 mg/ day), ibuprofen (200-800 mg/day), or celecoxib (200-300 mg/day) alone or in combination. The pre-and postoperative NLR was calculated from laboratory results obtained within 2 weeks before surgery and on postoperative day 1. The NLR of subsequent postoperative days was not calculated because routine blood specimens were not collected beyond postoperative day 1 in all the patients included in the analysis.

Statistical analysis and endpoints
The primary end points were RFS and OS (OS). RFS was defined as the time in months from the date of surgery to either the date of first evidence of recurrence or the date of death, whichever happened first. OS was defined as the time in months from the date of surgery to the date of death. Patients who were lost to follow up, remained disease-free, or were alive at the time of analysis were censored at the last known follow-up date.
Descriptive statistics for demographics and baseline patient characteristics included mean and standard deviation (SD) for continuous variables, while frequency counts and percentages were used for categorical variables. Fisher's exact test or chi-square test was used to evaluate the association between two categorical variables. Kruskal-Wallis test was used to evaluate the difference in a continuous variable between patient groups. Wilcoxon signed-rank test was used to compare the pre-and postoperative change in continuous variables. A cut-off value of 5 was used to discriminate between patients with high and low preoperative NLR (NLR ≥5 or <5) [8]. RFS and OS were analyzed using the Kaplan-Meier method and log-rank tests were used to assess the difference in timeto-event outcomes between patient groups. The median time-to-event in months and survival rates at 3 and 5 years were calculated including 95% confidence intervals. Univariate Cox proportional hazard models were used to evaluate differences in OS and RFS for continuous variables. Multivariable Cox proportional hazard models were used for multivariate analysis to include covariates that have shown to be associated with prognosis in previous studies or showed a P < 0.2 in the univariate analysis. P < 0.05 were considered statistically significant. Statistical software SAS 9.1.3 (SAS, Cary, NC) and S-Plus 8.0 (TIBCO Software Inc., Palo Alto, CA) were used for all the analyses.
The average preoperative NLR was 2.95 (SD = 2.43) and increased postoperatively to 10.28 (SD = 6.74) (P < 0.0001). There was a significant difference in preoperative NLR values and tumor stage with stage II having the highest NLR (mean = 3.25, SD = 2.34), followed by stage III (mean = 3.11, SD = 3.37) and stage I (mean = 2.75, SD = 1.86). One hundred and ten (9.66%) patients had a preoperative NLR ≥5. There was a trend toward a statistically significant difference between preoperative NLR group and tumor stage so that the patients with a high NLR (≥5) had higher tumor stage (stage I = 7.8% vs. stage II = 11.9% and stage III = 11.8%) (P = 0.07).
In regard to the perioperative use of NSAIDs, 563 (49.4%) patients were administered an NSAID. Three hundred and seventy-four of them received ketorolac alone or in combination with ibuprofen and/or celecoxib (Table 1). There were no statistically significant differences between the use of NSAIDs among patients with different tumor stages (P = 0.436). The average preoperative NLR of patients treated with NSAIDs (mean = 2.89, SD = 2.11) was not statistically different to that of patients who did not received NSAIDs (mean = 3.01, SD = 2.7, P = 0.259). Also, the proportion of patients with a preoperative NLR higher than 5 who were treated with NSAIDs (n = 55, 9.8%) or not (n = 55, 9.5%) was nearly identical (P = 0.899). Remarkably, the NSAIDs use were associated with a decrease in the rate of patients with a NLR higher than 5. Briefly, 79.4% (n = 447) of the patients with a NLR higher than 5 belonged to NSA-IDs group compared to 84.7% (n = 488, P = 0.191).
Although the average postoperative NLR was lower in patients who received NSAIDs (mean = 9.98, SD = 6.64) compared to those who did not receive the treatment (mean = 10.58, SD = 6.82), the difference was not statistically significant (P = 0.060).
The effect of preoperative NLR on RFS in patients with stage I NSCLC was statistically significant in the univariate Cox proportion analysis (P < 0.0001) but was not for patients with stage II (P = 0.51) or stage III (P = 0.56) NSCLC (Fig. 1). After adjustment for age, ASA 3-4, NSA-IDs use, neoadjuvant chemotherapy, adjuvant radiation, thoracoscopy, and postoperative NLR in the multivariate Cox proportional hazard model, the analysis demonstrated that a preoperative NLR ≥5 was an independent predictor of poor RFS patients in patients with stage I disease (HR = 2.13, 95% CI = 1.42-3.19, P = 0.0002) ( Table 4).
For the overall population of patients (n = 1139), postoperative NSAID administration was not predictive of RFS (any NSAID: P = 0.66, ketorolac: P = 0.48, celecoxib: P = 0.34, ibuprofen P = 0.86). When data were analyzed by tumor stage, postoperative use of NSAIDs was not associated with longer RFS (stage I, P = 0.12; stage II,  Univariate analysis demonstrated that only patients with stage I cancer who had a preoperative NLR ≥5 showed a worse OS survival than those with a lower than 5 NLR (P < 0.0001) (Fig. 1). The multivariate Cox proportional hazard model showed that the effect of a preoperative NLR ≥5 on OS was significant after adjusting for age, gender, ASA 3-4, NSAIDs use, adjuvant chemotherapy and radiation, thoracoscopy, and postoperative NLR (HR = 2.72; 95% CI = 1.75-4.23; P < 0.0001) ( Table 5). Overall, NSAID, celecoxib or ibuprofen, administration was not predictive of OS (NSAID: P = 0.18, celecoxib P = 0.19, ibuprofen P = 0.77) when all stages were included in the analysis. However, patients who were administered ketorolac demonstrated a marginally better OS (P = 0.05) in univariate analysis. When the analysis was conducted by tumor stage, univariate analysis demonstrated longer OS only for stage I patients who received NSAIDs compared with those not treated with these analgesics (P = 0.03) (Fig. 1). This association was not significant when adjusting for covariates including age, gender, ASA 3-4, adjuvant chemotherapy and radiation, type of surgery, and pre-and postoperative NLR (HR = 0.91; 95% CI = 0.65-1.26; P = 0.58) ( Table 6). The data analysis also demonstrated that the postoperative NLR was not a predictor of OS in patients who did (P = 0.64) or did not (P = 0.706) receive NSAIDs.

Discussion
This study demonstrates that a high preoperative NLR (≥5), a marker of inflammation, is associated with a   [3,14]. Others have also shown similar results; however, in those studies the cut-off NLR was 2.5, which makes difficult the comparison with our results [15,16]. Tumor stage might influence perioperative inflammation and immune suppression; therefore, we further analyzed the data to determine whether an NLR ≥5 was also a predictor of survival when considering tumor stage separately. A high preoperative NLR (≥ 5) was shown to be a significant independent predictor of decreased RFS and OS independent in patients with stage I disease but not stage II or III. Consistent with the findings reported by Sarraf et al., we found that the mean NLR was slightly, but statistically significantly higher in patients with tumor stage II and III than those in stage I, indicating a higher inflammatory status and perhaps more immune suppression in patients with more advanced stages of NSCLC [14]. This suggests, at least in our patient population, that preoperative immune suppression may be more important in patients with limited tumor burden. In addition, patients with higher stages are more likely to have longer, complex, and more invasive surgeries which are perioperative factors associated with a greater stress and inflammatory response [17,18]. Therefore, it could be argued that our results may reflect the association between type of surgery, early recovery, and improved survival; although this is merely speculative because there was only a trend toward NLR ≥5 in patients with stage III disease (P = 0.084).
NSAIDs are commonly given intra-and postoperatively as adjuvant analgesics. It has been suggested that the perioperative use of NSAIDs could be associated with prolonged RFS and OS [19,20]. Several mechanisms have been proposed including modulation of the inflammatory response and a direct positive effect on the tumor microenvironment and invasiveness of cancer cells [12]. Forget et al. demonstrated that intraoperative NSAID use was an independent predictor of increased distant metastasis-free survival, especially when given at the beginning of surgery, but not locoregional recurrence in patients with breast cancer [20]. Additionally, the study found that intraoperative ketorolac administration was an independent predictor of increased OS. In our work, we found that NSAID administration within 72 h after surgery was not predictive of longer OS. However, postoperative ketorolac administration (alone or in combination with other NSAIDS) showed a trend toward a significant association with an increase in OS but not RFS. These findings support previous studies suggesting that the addition of ketorolac, specifically, may have a positive prognostic effect on long-term survival [20,21].
Our study has several limitations, therefore, the results should be interpreted with caution. First, this is a retrospective study; hence there are unknown factors that might have affected the studied outcomes. Second, the  impact of immune suppression, inflammation, and effect of NSAIDs could be confounded by the histological type of tumor, type of recurrence (locoregional recurrence vs. distant metastasis), and adjuvant treatment. Moreover, the total amount of NSAIDs administered postoperatively was not quantified and this could potentially explain the discordant findings between the study by Forget et al. and our work. Third, the statistical power was limited by the number of patients available in our database and therefore there is a possibility of a type II error. A propensity score based on logistic regression was not conducted because the number of covariates included in the logistic regression model is generally calculated as the number of patients in the smallest group divided by 10 [22]. Therefore, with few patients per group by stage, we could put only four covariates in the model when in fact there were several covariates. Finally, the generalizability of our findings might be limited because our patient population was extracted from a large tertiary care only dedicated to cancer care.
In conclusion, a high preoperative NLR (≥5) was predictive of longer RFS and OS following resection in patients with stage I NSCLC. NSAID administration was not found to be an independent predictor of survival, however, postoperative ketorolac administration showed a trend toward better OS. Despite the relatively small number of patients receiving NSAIDs and the retrospective nature of this study, the results of this study provide support for future prospective studies comparing the effect of tumor stage, perioperative inflammatory status, and perioperative NSAIDs.