nlr-io

Code and citation for the finding that change in the neutrophil-to-lymphocyte ratio (NLR) during immune checkpoint inhibitor treatment is a non-linear predictor of overall survival in advanced cancer.

Citation

Li M, Spakowicz D, Burkart J, Patel S, Husain M, He K, Bertino EM, Shields PG, Carbone DP, Verschraegen CF, Presley CJ, Otterson GA, Kendra K, Owen DH. Change in neutrophil to lymphocyte ratio during immunotherapy treatment is a non-linear predictor of patient outcomes in advanced cancers. J Cancer Res Clin Oncol. 2019;145(10):2541–2546. doi:10.1007/s00432-019-02982-4. PMID:31367835. PMCID:PMC6751277.

@article{li2019nlr,
  title   = {Change in neutrophil to lymphocyte ratio during immunotherapy treatment is a non-linear predictor of patient outcomes in advanced cancers},
  author  = {Li, Mingjia and Spakowicz, Daniel and Burkart, Jarred and Patel, Sandip and Husain, Marium and He, Kai and Bertino, Erin M. and Shields, Peter G. and Carbone, David P. and Verschraegen, Claire F. and Presley, Carolyn J. and Otterson, Gregory A. and Kendra, Kari and Owen, Dwight H.},
  journal = {Journal of Cancer Research and Clinical Oncology},
  year    = {2019},
  volume  = {145},
  number  = {10},
  pages   = {2541--2546},
  doi     = {10.1007/s00432-019-02982-4},
  pmid    = {31367835},
  pmcid   = {PMC6751277}
}

GitHub can also format this citation from CITATION.cff.

What the paper found

NLR (blood neutrophils divided by lymphocytes) was already known to be prognostic at a single baseline cutpoint. This study measured it at the start of immunotherapy and again a median of 21 days later in 509 patients with advanced cancer treated at Ohio State University from 2011 to 2017.

• Baseline NLR below 5: median overall survival 20.7 months, versus 7.9 months at or above 5.
• On-treatment NLR below 5: 23.9 months, versus 7.1 months at or above 5.
• A moderate decrease in NLR (16.7–28.5%) had the longest survival, 27.8 months. A larger decrease (beyond 28.5%) was 11.4 months. An increase of more than 100% was 5.0 months.
• A cubic term for the change in NLR improved a Cox model and remained significant after adjustment for age, body mass index, sex, cancer type, performance status, and days to the repeat NLR.

The curves on the hex sticker and graphical abstract are schematics drawn from these published summary numbers.

Scripts

Script	What it does
fig2_change-in-nlr.Rmd	Reproduces Figure 2: the Kaplan–Meier split on change in NLR, the loess smooth of change versus follow-up time, and the Cox model with a cubic term. Writes PDFs to figures/.
images/render-readme-figures.R	Draws the hex sticker and the graphical abstract from the published summary numbers. Uses base R only.

fig2_change-in-nlr.Rmd expects nlr-survival-deident.RDS in the repository root, with columns months, vitalstatus (1 = death), nlr_baseline, and nlr_change.

# Figure 2 (requires the dataset below)
rmarkdown::render("fig2_change-in-nlr.Rmd")

# README figures
Rscript images/render-readme-figures.R

Figure 2 needs dplyr, ggplot2, survival, and survminer.

Data

The de-identified dataset nlr-survival-deident.RDS is available on request to academic researchers. Email daniel.spakowicz@osumc.edu from an institutional address.

License

MIT. See LICENSE.