Challenges and caveats of a multi-center retrospective radiomics study: an example of early treatment response assessment for NSCLC patients using FDG-PET/CT radiomics.

BACKGROUND: Prognostic models based on individual patient characteristics can improve treatment decisions and outcome in the future. In many (radiomic) studies, small size and heterogeneity of datasets is a challenge that often limits performance and potential clinical applicability of these models. The current study is example of a retrospective multi-centric study with challenges and caveats. To highlight common issues and emphasize potential pitfalls, we aimed for an extensive analysis of these multi-center pre-treatment datasets, with an additional 18F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) scan acquired during treatment. METHODS: The dataset consisted of 138 stage II-IV non-small cell lung cancer (NSCLC) patients from four different cohorts acquired from three different institutes. The differences between the cohorts were compared in terms of clinical characteristics and using the so-called 'cohort differences model' approach. Moreover, the potential prognostic performances for overall survival of radiomic features extracted from CT or FDG-PET, or relative or absolute differences between the scans at the two time points, were assessed using the LASSO regression method. Furthermore, the performances of five different classifiers were evaluated for all image sets. RESULTS: The individual cohorts substantially differed in terms of patient characteristics. Moreover, the cohort differences model indicated statistically significant differences between the cohorts. Neither LASSO nor any of the tested classifiers resulted in a clinical relevant prognostic model that could be validated on the available datasets. CONCLUSION: The results imply that the study might have been influenced by a limited sample size, heterogeneous patient characteristics, and inconsistent imaging parameters. No prognostic performance of FDG-PET or CT based radiomics models can be reported. This study highlights the necessity of extensive evaluations of cohorts and of validation datasets, especially in retrospective multi-centric datasets.

Phase II Study of a Radiotherapy Total Dose Increase in Hypoxic Lesions Identified by 18F-Misonidazole PET/CT in Patients with Non-Small Cell Lung Carcinoma (RTEP5 Study).

See an invited perspective on this article on page 1043.This multicenter phase II study investigated a selective radiotherapy dose increase to tumor areas with significant 18F-misonidazole (18F-FMISO) uptake in patients with non-small cell lung carcinoma (NSCLC). Methods: Eligible patients had locally advanced NSCLC and no contraindication to concomitant chemoradiotherapy. The 18F-FMISO uptake on PET/CT was assessed by trained experts. If there was no uptake, 66 Gy were delivered. In 18F-FMISO-positive patients, the contours of the hypoxic area were transferred to the radiation oncologist. It was necessary for the radiotherapy dose to be as high as possible while fulfilling dose-limiting constraints for the spinal cord and lungs. The primary endpoint was tumor response (complete response plus partial response) at 3 mo. The secondary endpoints were toxicity, disease-free survival (DFS), and overall survival at 1 y. The target sample size was set to demonstrate a response rate of 40% or more (bilateral α = 0.05, power 1-ß = 0.95). Results: Seventy-nine patients were preincluded, 54 were included, and 34 were 18F-FMISO-positive, 24 of whom received escalated doses of up to 86 Gy. The response rate at 3 mo was 31 of 54 (57%; 95% confidence interval [CI], 43%-71%) using RECIST 1.1 (17/34 responders in the 18F-FMISO-positive group). DFS and overall survival at 1 y were 0.86 (95% CI, 0.77-0.96) and 0.63 (95% CI, 0.49-0.74), respectively. DFS was longer in the 18F-FMISO-negative patients (P = 0.004). The radiotherapy dose was not associated with DFS when adjusting for the 18F-FMISO status. One toxic death (66 Gy) and 1 case of grade 4 pneumonitis (>66 Gy) were reported. Conclusion: Our approach results in a response rate of 40% or more, with acceptable toxicity. 18F-FMISO uptake in NSCLC patients is strongly associated with poor prognosis features that could not be reversed by radiotherapy doses up to 86 Gy.

Use of Metabolic Parameters as Prognostic Factors During Concomitant Chemoradiotherapy for Locally Advanced Cervical Cancer.

OBJECTIVES: To investigate the use of metabolic parameters as early prognostic factors during concomitant chemoradiotherapy for locally advanced cervix carcinoma (LACC). MATERIALS AND METHODS: Between February 2008 and January 2012, 34 consecutive patients treated for LACC (International Federation of Gynecology and Obstetrics Staging System stage IB2-IVA) were included in a retrospective study. Treatment was standard of care: total dose of 45 Gy in 1.8 Gy per fraction with concurrent cisplatin followed by brachytherapy. 18F-FDG PET-CT modalities were performed before treatment and per-treatment (at 40 Gy). The analyzed parameters were: maximum standardized uptake value (SUVmax), SUVmax variations of the primary tumor between the 2 investigations (DSUVmax), metabolic tumor volume (MTV), and total lesion glycolysis (TLG). Survival was assessed according to early metabolic changes during chemoradiotherapy. RESULTS: Median follow-up was 16 months (range, 5.3 to 32.4 mo). Median SUVmax before treatment was 13.15 (5.9 to 31) and was 5.05 (0 to 12) per-treatment. Median DSUVmax was 63.97% (0% to 100%). Median MTV before treatment was 44.16 mL (3.392 to 252.768 mL) and was 5.44 mL (0 to 69.88 mL) per-treatment. Median TLG before treatment was 249.82 mL (13.40 to 1931.10 mL) and was 20.14 mL (0 to 349.99 mL) per-treatment. At 40 Gy, SUVmax≥6, DSUVmax≤40%, MTV≥5.6 mL, and TLG≥21.6 mL were significantly associated with overall survival and progression-free survival reduction. MTV predicted progression with a sensitivity of 80% and a specificity of 87.5% and TLG with a sensitivity of 80% and a specificity of 83.3%. CONCLUSIONS: PET-CT imaging could be useful as an early prognostic factor during treatment for LACC. MTV and TLG seem to provide better prognostic information than SUVmax and DSUVmax.

Early assessment of metabolic response by 18F-FDG PET during concomitant radiochemotherapy of non-small cell lung carcinoma is associated with survival: a retrospective single-center study.

BACKGROUND AND PURPOSE: We performed a retrospective single-center study to assess if midtreatment 18F-FDG PET/CT could predict local control and survival in patients with locally advanced non-small cell lung cancer treated with concurrent chemoradiotherapy. METHODS: Thirty-one consecutive patients with unresectable or locally advanced lung cancer (T2-4 N0-3 M0) were treated with concurrent chemoradiotherapy in our center. Each patient received 18F-FDG PET/CT before treatment and at midtreatment time when a radiation therapy dose of 30 Gy was delivered. We assessed several PET/CT parameters as follows: SUV max, ΔSUV mean, ΔSUV max, variation of hypermetabolic tumor volume, and the variation of tumor total lesion glycolysis (ΔTLG). Univariate analysis was performed, and a stepwise procedure was used to define final multivariate model. RESULTS: The ΔTLG was statistically correlated to overall survival (OS) (P = 0.035), progression-free survival (P = 0.023), and local control (P = 0.043) in univariate analysis. A decrease in TLG over 15% was statistically correlated to a better OS (P = 0.007; hazards ratio [HR], 7.439; 95% confidence interval [CI], 1.168-28.897) and progression-free survival (P = 0.010; HR, 5.695; 95% CI, 1.506-21.537) in univariate analysis. In multivariate analysis, ΔTLG superior to -15% was significantly correlated to a worse OS (P = 0.020; HR, 5.973; 95% CI, 1.324-26.953). CONCLUSIONS: Early assessment of TLG response by 18F-FDG PET/CT during concomitant radiochemotherapy of non-small cell lung cancer might be associated with survival.

Prognostic value of an increase in the serum thyroglobulin level at the time of the first ablative radioiodine treatment in patients with differentiated thyroid cancer.

PURPOSE: The aim of this study was to assess the prognostic factors for treatment efficacy, and in particular the increase in serum thyroglobulin (Tg) level at the time of the first ablative radioiodine treatment, in patients with differentiated thyroid carcinoma (DTC). METHODS: A retrospective chart review was performed on 407 patients treated for DTC by total thyroidectomy and (131)I ablation between 1995 and 2002, and examined 5-12 months later with diagnostic (131)I whole-body scan and serum Tg measurement after thyroid hormone treatment withdrawal. At the time of the ablative radioiodine treatment, serum Tg level was determined just before (131)I administration (TgD0) and 5 days later (TgD5); Tg variation was expressed as the ratio TgD5/TgD0. At the first post-ablation follow-up examination, unsuccessful ablation was defined by a Tg level > or =2 ng/ml and/or abnormal (131)I uptake. RESULTS: Ablation was unsuccessful in 51 patients. Univariate analysis showed high TgD0 level, low TgD5/TgD0 ratio extrathyroidal invasion, (131)I uptake in the neck (excluding the thyroid bed) during the ablative treatment and distant metastases to be significantly associated with unsuccessful ablation. On logistic multivariate analysis, TgD0 level <5 ng/ml and TgD5/TgD0 ratio > or =20 were independently associated with successful ablation. A receiver operating characteristic curve analysis determined that a TgD5/TgD0 ratio greater than 20 had a 97% positive predictive value for successful ablation. When both TgD0 and TgD5/TgD0 ratio were considered, that is, TgD0 <5 ng/ml or TgD0 > or =5 ng/ml but TgD5/TgD0 ratio >20, ablation was unsuccessful in only 12/301 patients. CONCLUSION: Our data show that the TgD5/TgD0 ratio may be used as a new prognostic indicator of (131)I treatment efficacy in patients with DTC.