Abstract
Objectives
Baseline neutrophil–lymphocyte ratio (NLR) has been reported to be a prognostic biomarker for survival and a predictive marker for response to treatment in prostate cancer.1–3 It also has been hypothesised that ‘in patients with a high NLR, treatment with chemotherapy might be more beneficial than Abiraterone’.1,3 This hypothesis is based on retrospective analysis of data from phase III clinical trials, and hence the relevance and applicability to a ‘real world’ population is unknown. It is widely accepted that patients in clinical trials are physically more fit, well-motivated and have significantly less comorbidity due to restrictive trial eligibility criteria.4,5 We analysed whether NLR is prognostic and predictive in an unselected group of patients.
Design
Retrospective analysis of routinely collected patient data from 149 patients who were consecutively treated, with Abiraterone 1 g and Prednisolone 10 mg combination, between January 2012 and May 2014.
Setting
Nottingham University Hospital NHS Trust, Nottingham, UK.
Participants
Median age was 75 years (55 to 88 years); Baseline median PSA was 52.3 ng/ml (0.03 to 6884); Median haemoglobin was 12.4 g/dL (7.6 to 16.0). Median alkaline phosphatase was 109 (38 to 2728). Bone metastasis, Nodal metastasis and Visceral metastasis were present in 90%, 38%, and 14%, respectively. Ninety-four patients (63%) were Docetaxel naive.
Main outcome measures
The prognostic and predictive value of baseline neutrophil–lymphocyte ratio was assessed.
Results
Baseline median NLR was 3.05 (range 0.8 to 32.4) and mean NLR 4.58 (95% CI 3.82 to 5.33). The overall PSA 50% response rate was 49%. Haemoglobin >11.0 was associated with better PSA response with an odds ratio of 2.4 (95% CI 1.2 to 4.8). Chemotherapy naive patients, as expected, also had a better response to Abiraterone with an odds ratio of 3.1 (95% CI 1.4 to 7.3).
A high NLR >4 was confirmed to be prognostic and was associated with inferior overall survival (1-year survival rate 59% vs 77%, log rank test: p = 0.054). But contrary to van Soest et al. 1 findings, NLR, as a dichotomised variable, was not predictive of treatment response at any ratio (NLR 2, 3, 4 or 5). For instance, with a NLR ratio of 4, the PSA 50% response rate was 55% versus 47% in high (<4) and low (<4) NLR groups, respectively (chi-square test p-value 0.32). Furthermore, NLR, as continuous variable, was also not predictive of treatment response (odds ratio 1.05 on logistic regression p 0.172).
There was no imbalance in patient characteristics such as age, baseline PSA, haemoglobin, ALP, sites of metastasis (bone, vertebral or nodes) between high (<4) and low (>4) NLR groups (p-value at least >0.1 on Mann-Whitney, chi-square or student t-test as appropriate).
More patients pre-treated with Docetaxel had a high NLR of >4 (48% vs 31.3% vs p = 0.046). But on subgroup analysis, NLR was not predictive of treatment response both in Docetaxel naive and Docetaxel exposed patients (Chi-square p 0.30 and p 0.784, respectively).
The discordant results could be explained by the fact that high NLR is a global and not a cancer specific biomarker. 6 High NLR has been associated with adverse outcomes in many common surgical and medical conditions. For instance, a high NLR is associated with increased cardiovascular disease mortality, smoking, diabetes, hypertension, hypercholesterolemia, ageing, body mass index, renal failure, metabolic syndrome, chronic obstructive pulmonary disease, and heart failure. 6
Conclusion
We conclude that in this non-trial population, where there is a higher prevalence of comorbid medical conditions, NLR, although prognostic, is not a useful predictive biomarker for optimising the sequencing of hormonal and chemotherapeutic agents in metastatic castration-resistant prostate cancer.