# [18F]FDG-PET/CT-based evaluation of tumor response kinetics during induction chemotherapy and concurrent radiochemotherapy in stage II - III small-cell lung cancer

## TL;DR

This study shows that interim [18F]FDG-PET/CT scans during chemotherapy can predict survival outcomes in small-cell lung cancer patients.

## Contribution

A novel exponential decay model was developed to evaluate tumor response kinetics using PET/CT scans at variable time points.

## Key findings

- Poor SUVmax or TLG response was significantly associated with shorter progression-free survival.
- Interim PET/CT response parameters can guide individualized treatment adaptation.
- SUVmax showed greater inter-patient variability compared to MTV.

## Abstract

Evaluation of interim-[18F]FDG-PET/CT as a prognostic tool in limited disease small cell lung cancer (SCLC).

We included 35 patients with limited disease SCLC from a prospective institutional registry in this retrospective study. Patients received induction chemotherapy (3–4 cycles) followed by concurrent radiochemotherapy. Baseline [18F]FDG-PET/CT was performed before or shortly after start of induction chemotherapy, interim PET/CT was acquired during late induction or concurrent chemoradiotherapy. Maximum standardized uptake value (SUVmax), metabolic target volume (MTV), and total lesion glycolysis values (TLG) were determined. An exponential decay model with an asymptotic offset was used to describe treatment response over time. Deviations > 2 standard deviations (SD) above model-predicted means after day 30 were considered poor response. Progression-free survival (PFS) was analyzed.

All patients underwent twice-daily radiotherapy to a base dose of 45 Gy. SUVmax showed greater inter-patient variability than MTV. Poor treatment response was observed in 17%, 31%, or 14% at the SUVmax, MTV and TLG endpoints. Deviations > 2 SD from the model in SUVmax and TLG were significantly associated with shorter PFS (p = 0.0003, p = 0.0014); MTV was not prognostic (p = 0.2630). Leave-one-out cross-validation (LOOCV) could confirm the prognostic value of the standardized residual SUVmax larger than 2 standard deviations above model estimate as negative PFS predictor (p = 0.0152, Fishers exact test).

Our decay model enables the characterization of [18F]FDG-PET/CT response parameters from scans acquired at variable time points during induction chemotherapy and at start of concurrent radiochemotherapy. Poor SUVmax or TLG response was predictive of PFS. Interim-[18F]FDG-PET/CT response may guide individualized treatment adaptation.