J. Price^1,2, H. Mistry², L. Goyal¹, L. Barraclough¹, S. Morrison¹, K. Haslett¹, D. Thomson^1,2, and C. M. L. West²; ¹The Christie NHS Foundation Trust, Manchester, United Kingdom, ²Division of Cancer Sciences, The University of Manchester, Manchester, United Kingdom

Purpose/Objective(s):

The host immune system affects radiotherapy (RT) response. For oropharynx carcinoma, we previously showed the prognostic significance of pre-RT absolute lymphocyte counts (ALCs) depends on whether patients receive concurrent cisplatin chemotherapy, suggesting a hitherto underappreciated immunomodulating effect of cisplatin. In this work, we hypothesized a similar interaction exists for patients with cervical cancer.

Materials/

Methods:

Our study was a retrospective analysis of prospectively collected data from a single institution. Inclusion criteria: non-metastatic cervical carcinoma treated with curative-intent (chemo)RT, Jan 2013 - Dec 2019. Exclusion criteria: neoadjuvant chemotherapy. Patient-, cancer- and treatment-factors were collected, along with baseline ALC values (from four weeks prior to RT, to RT start). The primary endpoint was overall survival (OS). ALC distributions were plotted and correlations with predictors assessed. ALC vs log(HR) was plotted using splines. A Cox proportional hazards (PH) model was fitted; predictors included log(ALC), concurrent cisplatin use, an interaction term, and other relevant predictors.

Results:

Patients (n=487) had a median age of 49 years (interquartile range [IQR] 39-62 years). 90% had an ECOG performance status 0-1; 45% were never smokers, 76% had squamous cell histology and 75% had concurrent cisplatin. 24%, 53%, 10% and 12% had FIGO stage I, II, III and IVa disease, respectively. The median pre-RT ALC was 1.8 (IQR: 1.3-2.2); the distribution of ALCs was not normal and so ALCs were log-transformed. The distribution of ALCs did not differ according to other predictors. The relationship of ALC vs log(HR) was u-shaped, with (log) hazard of death falling from 0.5 to <0 for ALCs of 0-1.5, before increasing above 0 for ALC values >3. Kaplan-Meier plots of estimated OS according to ALC (0-1.5 vs 1.5-3 vs >3) stratified for OS for patients receiving chemo-RT, but not those receiving RT alone. On Cox PH modelling, an interaction was identified between ALC and concurrent cisplatin use (likelihood ratio test [LRT] p<0.000001); the prognostic value of ALCs was dependent on chemotherapy use (and vice versa). The interaction persisted when adjusted for other relevant predictors (Table).

Conclusion:

We demonstrated an interaction between pre-RT ALC and concurrent cisplatin for patients with cervical cancer, adding to our prior work in oropharynx cancer, and to increasing evidence of an immuno-modulatory effect of cisplatin. ALC may represent a candidate biomarker to guide chemo-RT: immunotherapy combinations. Further pre-clinical work may elucidate an underlying mechanism.

Abstract 3553 – Table 1

Characteristic	HR1	95% CI1	p-value
Age	1.01	1.00, 1.02	0.063
FIGO stage
I	—	—
II	2.25	1.35, 3.74	0.002
III	5.91	3.25, 10.7	<0.001
IV	5.96	3.43, 10.3	<0.001
ECOG PS
0	—	—
1	1.81	1.27, 2.58	<0.001
2	2.10	1.33, 3.31	0.001
3	3.30	1.17, 9.26	0.023
Log(ALC)	0.71	0.49, 1.02	0.065
Log(ALC) * conc_chemo
Log(ALC) * No	2.13	1.27, 3.58	0.004
1HR = Hazard Ratio, CI = Confidence Interval