GAPDH is a predictor of response to R-CHOP in DLBCL

On 28 February 2019, Johanna Chiche from Université Côte d’Azur, Nice, FR and colleagues, published in Cell Metabolism the results of a study investigating potential prognostic factors of response in diffuse large B-cell (DLBCL) patients.

The authors also sought to identify further predictors of response to metabolic inhibitors in DLBCL patient samples. The molecular pathways involved in the identified prognostic factors from the DLBCL patient dataset were assessed in vitro and in vivo with the help of mouse primary lymphoma cell lines and transgenic mouse models, respectively.

Study design

• Gene expression patient dataset: N = 414 newly-diagnosed DLBCL patients treated with either cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP; n = 181) or with rituximab plus CHOP (R-CHOP; n = 233)
• Immunohistochemistry patient dataset: N = 43 newly-diagnosed DLBCL patient samples (training cohort) & N = 294 de novo DLBCL patient samples treated with R-CHOP (validation cohort)
• Primary mouse lymphoma lines: Eµ-Myc mouse primary lymphoma cell line
• Mouse lymphoma models: C57BL/6 Eµ-Myc transgenic mice that spontaneously develops non-Hodgkin B-cell lymphoma
• KTM therapy: N = 4 DLBCL patients refractory to all R-based therapies
  • Dosing: four-week cycles with coli L-asparaginase (K; on Days 1, 3, 5, 7, 9 , 11, and 13), mTOR inhibitor temsirolimus (T; on Days 1, 7, and 14) and metformin (M; in the last two weeks instead of K)

Key results

• Gene expression profiling of the R-CHOP DLBCL patient dataset (n = 233) identified 524 unique genes associated with differences in overall survival (OS) and R-CHOP treatment. Among those:
  • High expression levels of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was a predictor for favourable outcome (three independent probesets: HR = 0.53, P = 0.01573; HR = 0.54, P = 0.01771; HR = 0.53, P = 0.01929)
  • Lactate dehydrogenase b subunit (lDH-B) was significantly associated with a poor outcome
• DLBCL patients with low GAPDH levels (protein or mRNA) had a lower OS upon R-CHOP than those with high GAPDH levels (protein or mRNA; P = 0.0091 for protein levels)
• GAPDH expression was not associated with age-adjusted International Prognostic Index (aaIPI), cell-of-origin (COO), Hans classification or with expression of already established bad prognosis markers (Bcl2, Bcl6, Myc) and translocations (double- or triple-hit)
• Multivariate analysis further confirmed that high expression of GAPDH is a significant predictor of improved OS (HR = 0.603; P = 0.0371)
• In vitro primary mouse lymphoma cell lines and in vivo lymphoma mouse models indicated that:
  • Mouse primary lymphoma B-cells that express low levels of GAPDH were sensitive to mitochondrial electron transport chain (ETC) Complex I inhibition
  • Most likely the mTORC1 signalling pathway is involved in the regulation of cellular metabolism in mouse primary lymphoma B-cells that express low GAPDH
  • Mouse primary B-cell lymphoma cells expressing low GAPDH levels rely on glutamine metabolism
  • GAPDH expression influences the metabolic state and the sensitivity of mouse primary lymphoma B-cells to phenformin or coli L-asparaginase
• Based on all the in vitro and in vivo evidence mentioned above the investigators designed and trialled a new regimen (KTM) to interfere with the cellular metabolism seen with low expression levels of GAPDH:
  • Three out of four patients who received KTM achieved a complete response after two cycles of treatment
  • One patient had progressive disease due to early toxicity during the first treatment cycle
  • Median duration of response (range): 6 (4–6) months

Conclusions

The results of this study indicate that GAPDH is a predictor of DLBCL outcomes upon R-CHOP treatment. Patients with low GAPDH expression levels were low responders to R-CHOP and have a lower OS. These low GAPDH levels were associated with oxidative phosphorylation cellular metabolism, mTORC1 signalling, and glutaminolysis. When these cellular pathways were inhibited clinically (specific metabolic inhibitors) in DLBCL patients with low GAPDH levels and in lymphoma mouse models, responses were improved.