Combined impact of lipidomic and genetic aberrations on clinical outcomes in metastatic castration-resistant prostate cancer

Prostate cancer (PC) is the second most frequent cancer and fifth leading cause of cancer death in men worldwide [1]. Therapies such as taxanes, androgen receptor signalling inhibitors (ARSI), poly-ADP ribose polymerase inhibitors (PARPi) and targeted radioisotopes have significantly increased survival in metastatic castration-resistant prostate cancer (mCRPC); however, treatments driven by molecular subtyping are currently limited to PARPi in the setting of DNA repair gene aberrations and pembrolizumab in men with microsatellite instability [2]. Moreover, long-term control of mCRPC requires a combined approach targeting multiple hallmarks of cancer, encompassing the cancer genome, the immune system and metabolic factors including lipid metabolism, all of which contribute to cancer progression and treatment resistance [3].

The genomic landscape of mCRPC is well characterised [4], with numerous somatic genetic aberrations linked to poor clinical outcomes. Androgen receptor (AR) aberrations (50–70%), TP53 aberrations (50%), PTEN deletion (30%) and RB1 deletion (20%) are all associated with shorter time on ARSI [4‐7]. RB1 deletion is also correlated with shorter overall survival (OS) in mCRPC [6]. Aggressive-variant prostate cancer (AVPC), a subset of mCRPC which is rapidly progressive with a poor prognosis, is morphologically heterogenous, but has been shown to correlate with genetic aberrations in two or more of TP53, RB1 and/or PTEN [8].

Altered lipid metabolism and its impact on PC is increasingly recognised through epidemiological [9‐11] and molecular [12, 13] studies. Lipidomic profiling of plasma from men with PC has demonstrated that the plasma lipid profiles including elevated sphingolipids are associated with a poorer prognosis [14‐17]. Ceramide, a key sphingolipid, can be metabolised into sphingosine-1-phosphate (S1P) or sphingomyelin to promote cancer cell growth, tumour metastasis and treatment resistance [18‐20]. Elevated circulating sphingolipids, in particular ceramides, are associated with poorer clinical outcomes across the natural history of PC, including metastatic relapse in localised PC, earlier androgen deprivation failure in metastatic hormone-sensitive PC, and shorter OS and radiographic progression-free survival (rPFS) in mCRPC [14‐17]. We have derived and validated a poor prognostic 3-lipid signature (3LS) consisting of ceramide(d18:1/24:1), sphingomyelin(d18:2/16:0) and phosphatidylcholine(16:0/16:0) to represent the poor prognostic lipidomic profile [14]. The 3LS is independently associated with shorter OS in men with mCRPC commencing docetaxel in internal and external validation cohorts [14, 15].

A key question arises from lipidomic profiling studies—how do circulating lipid aberrations relate to the genomic landscape of prostate cancer? Genetic alterations can lead to metabolic reprogramming in cancer, and conversely, metabolic dysregulation can be an initiator of malignant cellular transformation [21]. The current study is uniquely placed to address this question and provide insights into new potential therapeutic strategies due to the multidisciplinary approach of parallel metabolic and genomic profiling. Thus, we aimed to assess the relationship between the poor prognostic 3-lipid signature, somatic genetic aberrations and clinical outcomes in mCRPC.

This study provides new insights into the correlation between circulating lipids and tumour genotypes in mCRPC. Elevated circulating sphingolipids were correlated with AR, RB1 and PI3K aberrations in two independent cohorts and associated with TP53 aberrations in the discovery cohort, but not with DNA repair, MMR or WNT pathway aberrations. The presence of a somatic aberration (AR, TP53, RB1, PI3K) and/or the 3LS was independently associated with shorter OS in multivariable analysis in both cohorts. In addition, AVPC was associated with elevated sphingolipids and the combination of AVPC and 3LS predicted for a shorter median survival of ~12 months. Taken together, these findings provide the first understanding into the correlation between circulating lipid and somatic gene alterations in mCRPC and suggest that men with AR, TP53, RB1, PI3K and AVPC aberrations are more likely to benefit from therapies targeting sphingolipid metabolism than men with DNA repair, MMR or WNT pathway anomalies.

A key question arising from this data is how the genomic and lipidomic changes are related biologically. Not all genotypes are associated with the sphingolipid profiles, suggesting that certain genetic aberrations (AR, TP53, RB1, PI3K) are more likely to be dysregulating lipid metabolism. Bivariate analysis of the genotypes with the 3LS was inconclusive regarding whether the genetics was driving the lipid metabolism or vice versa (Additional File 1: Tables S9.1-9.4).

There is emerging evidence of a biological link between lipid metabolism and genetic aberrations including AR, RB1 and PI3K on PC growth. Enhanced ceramide-S1P signalling may be mediating ARSI resistance induced by AR gain, as men with mCRPC had significantly shorter ARSI treatment duration if their tumours had AR gain in combination with increased expression of sphingolipid genes (involved in ceramide-S1P signalling) [16]. Furthermore, de novo resistance to enzalutamide in androgen-independent cells can be reversed with sphingosine kinase inhibitors in vitro [16]. A recent study demonstrated that a novel fatty acid synthase inhibitor antagonised CRPC growth through metabolic reprogramming and inhibited expression of AR and its variants, including AR-V7 [28]. Another study found that 3-hydroxy-3-methyl-glutaryl–CoA reductase (HMGCR), a key enzyme in the cholesterol synthesis pathway, was elevated in enzalutamide-resistant PC cell lines and that simvastatin, a HMGCR inhibitor, blocked AR synthesis and inhibited growth in vitro and in vivo [29]. There is also recent evidence suggesting a role for RB1 in regulating sphingolipid metabolism in advanced PC. In an in vitro model of mCRPC with RB1 deletion utilising the androgen-independent PC cell line C4-2 after RB knockdown, 27% of upregulated genes were involved in metabolic pathways, including sphingolipid metabolism [30]. In addition, ceramide and its metabolite S1P play key roles in the PI3K pathway, as a negative regulator and activator respectively [31]. Overall, these studies suggest that lipid metabolism may play a biological and possibly clinically relevant role in some of these molecular pathways.

These findings raise the question of whether manipulation of the lipidome might be an effective therapeutic strategy in addition to existing drugs, particularly for patients with unfavourable lipidomic and somatic genetic aberrations. For example in the IPATential150 study, men with mCRPC with PTEN-loss tumours had improved rPFS on treatment with abiraterone and an AKT inhibitor versus abiraterone alone (18.5 months vs 16.5 months, p = 0.034) [32]. Our data supports the hypothesis that the addition of a metabolic modifier in men with both a PI3K aberration and 3LS may increase the efficacy of the AKT inhibitor, and this warrants further investigation.

Metabolic therapies are currently not the standard of care for PC, but are used in the treatment of other diseases such as cardiovascular disease and diabetes. Non-pharmacological interventions such as exercise and diet can reduce ceramides [33, 34]. The cholesterol lowering drugs statins and PCSK9 inhibitors reduce levels of circulating ceramides and other sphingolipids in patients with dyslipidaemia [35, 36]. Several studies have already indicated that statin therapy is associated with improved OS in PC [37‐40]; however, blood cholesterol levels were not associated with prognosis in prostate cancer [41‐43]. Therefore, the beneficial effects of statin therapy may be related to other lipids such as ceramides given that statin can lower the circulating levels of these lipids.

Drugs specifically targeting sphingolipid metabolism are mostly in pre-clinical development in cancer [18], but a specific inhibitor of sphingosine kinase 2, Opaganib (ABC294640), has been tested in patients with solid tumours [44]. Recently, we showed that Opaganib can overcome de novo enzalutamide resistance in androgen-independent PC cells in vitro [16]. Opaganib is currently undergoing a Phase 2 study in combination with AR inhibitors in patients with mCRPC (NCT04207255). Clinical trials are required to determine if repurposing statins or other sphingolipid-targeting therapies for prostate cancer in combination with standard of care can improve clinical outcomes.

Furthermore, plasma lipidomic signatures predicting for high-risk patients in cardiovascular disease already exist [45] and have been translated into a rapid-turnover plasma ceramide test by the Mayo Clinic Laboratories [46]. Overall, the availability of metabolic therapies that can target sphingolipid metabolism in combination with genotyping and lipidomic analysis for patient selection, demonstrates the feasibility of personalised metabolic therapy in a clinical setting for men with mCRPC.

A strength of this study is the analysis of two independent cohorts of prospectively enrolled men with mCRPC using the previously validated 3LS [14, 15], with similar observations in both the discovery and validation cohorts. This study was limited by exclusion of men without available cfDNA for sequencing (29% of discovery cohort, 34% of validation cohort). High levels of ctDNA are associated with poorer clinical outcomes and likely to represent a higher tumour burden [7]. Therefore, the exclusion of men with undetectable/low cfDNA from our cohorts may have resulted in the cohorts having worse clinical outcomes compared to the average population of men with mCRPC. Inadvertent skewing of the cohort characteristics might account for differences between the discovery and validation cohorts (e.g. less significant differences in sphingolipid levels between men with and without genetic aberrations in the validation cohort). The cohorts also had differences in baseline PSA levels and Gleason grade—the median PSA levels were two times higher in the discovery cohort (30 vs 14 ng/mL), and a higher proportion of the discovery cohort had Gleason Grade ≥9 (49% vs 32%). Overall, the relatively small sample size of the cohorts limits clinical applicability and will need to be addressed in the future with studies of larger cohorts. Another potential confounder was the presence of co-occurring genetic aberrations.

Elevated circulating sphingolipids, especially ceramides, were associated with AR, TP53, RB1, PI3K and AVPC aberrations in mCRPC, and the combination of lipid and genetic abnormalities conferred a worse prognosis. This suggests that approaches targeting the aberrant lipid metabolism defined by the 3LS should be considered in men with these mCRPC genotypes in prospective clinical trials.