Blood based genomic assessment of the clinical efficacy and toxicity of peptide receptor radionuclide therapy (PRRT)

Introduction: Peptide receptor radionuclide therapy (PRRT) is effective in neuroendocrine tumor (NET) management but there are limited tools for monitoring (RECIST) and prediction of tumor response, and for the delineation of side-effects. Accurate monitoring of tumor response is constrained by post-PRRT imaging limitations, and there are no effective blood-genomic assays for hematologic toxicity. We utilized three independent blood-based gene expression assays: a 51-marker gene NETest to monitor therapeutic efficacy, PRRT Predictor Quotient (PPQ) - a molecular marker used to predict PRRT responsiveness and a 16 gene radiation-toxicity (RAD-TOX assay) to assess PRRT-related toxicity.

Methods: MSKCC ¹⁷⁷Lu-PRRT-treated GEP-NET and lung cohort (n = 50; median age 62: range 29-86, M:F23:27). Tumors were pancreatic (n=24), small bowel NET (n=18), lung (n=8). Grade was G1/G2 (n=38), G3 (n=9). All were metastatic (88% liver) and had disease progression at start. Patients received 1-5 prior treatments (median 3) including somatostatin analogues (82%), surgery (52%) and chemotherapy (46%). PRRT treatment response assessed by radiographic report classified as: stable, partial response or progressive disease. Hematological radiation toxicity >Grade 2 for WBC, platelet and/or hemoglobin. Blood sampling: pre-PRRT and before cycle 3. Gene expression assays: RNA isolation, real-time qPCR and multi-algorithm analyses. NETest (0-100 score). RAD-TOX score (-100 to +100). PPQ (positive = predict “responder”/negative = predict “non-responder”). The RADTOX-gene signature was derived from genome-wide transcriptomic evaluations of radiation-toxicity, -sensitivity and -response gene expression studies (n=10,000 samples). Samples de-identified, assay and analyses blinded. Statistics: Mann-Whitney U-Test (2-tailed).

Results: Thirty-one patients (62%) exhibited response (unchanged/decreased disease). Progression (non-response) occurred in 17 (34%). PPQ was accurate in 98% (30/31 responders). In PRRT responders, NETest score decreased significantly (p = 0.03) before cycle 3 (-17 ± 9%); in "non-responders," NETest increased (+ 64 ± 15%) (p = 0.0003). Hematological toxicity (>Grade 2) developed in 24 (48%). The RAD-TOX scores significantly increased (+ 294 ± 55%) (p < 0.0001) in those who developed PRRT toxicity but were unchanged or decreased (- 105 ± 34%) (p = 0.023) in the non-toxic group. PPQ was overall accurate in 98% (30/31 responders).

Conclusions: The NETest can be used as an effective early monitor of PRRT-response. A NETest decrease during PRRT identified responders while an increase categorized non-responders. PPQ predicted PRRT response in 98%. A 16 gene-RAD-TOX panel accurately correlated with PRRT-induced hematological toxicity. Blood-based molecular gene signatures are promising noninvasive tools which can provide clinical guidance and enhance management during PRRT, particularly, since radiographic pseudo-progression is a known entity during the course of PRRT.