MTAP deficiency creates an exploitable target for antifolate therapy in 9p21-loss cancers

Abstract

Methylthioadenosine phosphorylase, an essential enzyme for the adenine salvage pathway, is often deficient (MTAP^def) in tumors with 9p21 loss and hypothetically renders tumors susceptible to synthetic lethality by antifolates targeting de novo purine synthesis. Here we report our single arm phase II trial (NCT02693717) that assesses pemetrexed in MTAP^def urothelial carcinoma (UC) with the primary endpoint of overall response rate (ORR). Three of 7 enrolled MTAP^def patients show response to pemetrexed (ORR 43%). Furthermore, a historic cohort shows 4 of 4 MTAP^def patients respond to pemetrexed as compared to 1 of 10 MTAP-proficient patients. In vitro and in vivo preclinical data using UC cell lines demonstrate increased sensitivity to pemetrexed by inducing DNA damage, and distorting nucleotide pools. In addition, MTAP-knockdown increases sensitivity to pemetrexed. Furthermore, in a lung adenocarcinoma retrospective cohort (N = 72) from the published BATTLE2 clinical trial (NCT01248247), MTAP^def associates with an improved response rate to pemetrexed. Our data demonstrate a synthetic lethal interaction between MTAP^def and de novo purine inhibition, which represents a promising therapeutic strategy for larger prospective trials.

Fig. 1. MTAP deficient metastatic urothelial carcinoma response to pemetrexed.

a schematic illustration of salvage and de novo adenine synthesis pathways in the context of MTAP loss and de novo purine synthesis inhibition. Inhibition of the de novo pathway leads to decreased nucleotide synthesis and eventually tumor cell apoptosis. MTA methylthioadenosine, PRPP phosphosphoribosyl pyrophosphate. b Frequency of MTAP deletion in 408 UC patients in The Cancer Genome Atlas (TCGA) database. HD, homozygous deletion; LOH, loss of heterozygosity; LLG, low-level copy number gain; Ampl, amplification. c, d Frequency of MTAP loss in a tumor tissue microarray. Positive and negative staining of MTAP within 109 MTAP^prof and 42 MTAP^def samples are available in Source Data file. Tis, carcinoma in situ; T1, invasive to lamina propria; T2-4a, muscle-invasive carcinoma; N + , metastatic to nodes; M + , systemic metastases. e Waterfall plots for best response of target lesions based on retrospective analysis of patients with metastatic UC treated with pemetrexed. f Clinical trial schema of NCT02693717. g Waterfall plots for best response of target lesions of patients with metastatic UC treated with pemetrexed under NCT02693717. *not evaluable for response; °progressive disease for new lesions. h Example of a patient with metastatic MTAP^def UC who progressed after receiving gemcitabine/cisplatin but responded to pemetrexed, compared to patient with MTAP^prof UC who progressed after receiving gemcitabine/cisplatin but didn’t respond to pemetrexed.

Fig. 2. In vitro and in vivo effects of pemetrexed on human UC based on MTAP protein status.

a Western blot verification of MTAP protein loss in four MTAP^def vs four MTAP^prof human UC cell lines paired with UHPLC-ESI/triple quadrupole mass spectrometry measuring MTA concentration in cell media of human UC cell lines (ng/mL). MTA levels are significantly higher in MTAP^def compared to MTAP^prof cell lines P value was calculated by Welch T test. b Sub-G1 analysis of MTAP^def and MTAP^prof cell lines when treated with 0.5 μM and 20 μM of pemetrexed (PEM). P value was calculated by Mann-Whitney test. c Cell viability after treatment with increasing concentrations of pemetrexed is significantly higher in MTAP^prof compared to MTAP^def cell lines P value was calculated by two-way ANOVA. d, e Xenograft tumor volume in HT-1376 and UM-UC-3 when treated with pemetrexed vs solvent. P value was calculated by two-way ANOVA. f Western blot and UHPLC-ESI/triple quadrupole mass spectrometry confirmation of MTAP knockdown in two HT-1376 cell lines: shMTAP2 and shMTAP3. g Pemetrexed resulted in significantly lower viability in the transfected HT1376 compared to the parental cell lines, with or without transduction of the lentivirus control vector. P value was calculated by two-way ANOVA. Data in a (bottom panel), b–e, f (bottom panel), and g are presented as mean + /− SD.

Fig. 3. Pemetrexed (PEM) induces significantly higher DNA damage in MTAP^def cell lines as compared to MTAP^prof cell lines.

Eight human bladder cancer cell lines were treated with no treatment (a) or 5 μM of PEM (b) for 24 h. Cells were then fixed and stained with γ-H2AX (red), and 53BP1 (green) antibodies as well as DAPI stain (blue). Images were quantified with ImageJ for γ-H2AX (c) and 53BP1 (d). Data represent mean ± SEM of four fields and analyzed with Welch T test.

Fig. 4. MTAP deficiency leads increased sensitivity to folate-based therapy in lung adenocarcinoma.

a Retrospective analysis schema for the BATTLE-2 trial. b Scatterplot of CDKN2A and MTAP RNA expression divided into four distinct groups with CDKN2A^lo/MTAP^lo and CDKN2A^hi/MTAP^hi having no overlap. MTAP cutoff value was 5.44 and CDKN2A cutoff value was 4.6 c Response rates to pemetrexed-based therapy in CDKN2A^lo/MTAP^lo vs all other groups. Difference is statistically significant by two-sided Fisher’s exact test (p = 0.0115). d Generalized linear regression model evaluating the correlation of 10 most altered genes in lung cancer beside MTAP to estimate the odds ratio and p value for each gene independently. Genes with an odds ratio >1 (log (odds ratio) >0) and a p value <0.05 are considered to be positively correlated with response. Genes with an odds ratio <1 (log (odds ratio) <0) and a p value <0.05 are considered to be negatively correlated with response. Adjustments were made for multiple gene comparisons and q value are presented in supplementary table S5. e Example of a patient with metastatic CDKN2A^lo/MTAP^lo lung adenocarcinoma who partially responded to pemetrexed-based therapy compared to a patient with CDKN2A^hi/MTAP^hi lung adenocarcimoma who progressed after pemetrexed-based therapy.