Genomic Analysis of Three Metastatic Prostate Cancer Patients with Exceptional Responses to Carboplatin Indicating Different Types of DNA Repair Deficiency

1. Introduction

A distinct subgroup of patients with metastatic castration-resistant prostate cancer (mCRPC) [1] carry germline or somatic defects in DNA repair genes. Some of these alterations associate with significant and long-lasting responses to poly (ADP-ribose) polymerase (PARP) inhibitors [2]. In particular, homologous recombination defects (HRD) are also known to confer sensitivity to platinum derivatives in diverse cancers [3], [4]. Nevertheless, platinum has not been routinely used for the care of advanced prostate cancer (PCa) as the only randomised phase III trial with a platinum compound in molecularly unselected mCRPC patients failed to show overall survival benefit [5]. Here we report on three patients with metastatic PCa with strong family histories and different DNA repair defects who have had remarkable and highly durable responses to platinum-based treatment and unequivocal clinical benefit. Interestingly, no specific DNA repair defects were detected in the first patient but whole exome sequencing revealed a mutational signature with a high homologous recombination repair deficiency score. In the second and third patient, deleterious germline BRCA2 and ATM mutations, respectively, were identified.

Patient 1 was aged 52 yr when he presented in January 2014 with adenocarcinoma of the prostate with extensive nodal, lung, and bone metastatic disease and a relatively low prostate-specific antigen (PSA: 8.8 ng/ml). His family history was highly suspicious for a cancer syndrome (Fig. 1). He was treated initially with androgen deprivation therapy (ADT) and radiotherapy (RT) to the thoracic spine but progressed after 4 mo with multiple liver metastases and worsening of the lung lesions. Docetaxel, cabazitaxel (four cycles each), and enzalutamide (two cycles) administered in sequence failed to provide a response. A liver biopsy at this time confirmed the presence of prostate adenocarcinoma (Fig. 2). Having exhausted all the standard treatment options, single-agent carboplatin was initiated at the target dose of area under the curve (AUC) 6 (total dose of 1100 mg) based on ethylenediamine tetra acetic acid. The first restaging after the second cycle showed a dramatic radiological response (70% reduction of the measurable disease) with further regression until cycle 11 when treatment was discontinued due to prolonged neutropenia. After 10 mo, he presented with symptomatic liver disease and a solitary cerebellar lesion. Re-challenge with carboplatin was once again successful, with regression of the cerebellar lesion indicating carboplatin delivery to the brain (Fig. 3). CyberKnife of his solitary brain lesion was administered after two cycles of carboplatin, which was resumed 2 mo after this RT, for four additional cycles but then again stopped due to persistent neutropenia. Restaging in October 2017, 2 mo after carboplatin interruption, indicated progression of the liver metastases and stable disease elsewhere. In February 2018, after participation in a phase I clinical trial, he was re-challenged for the third time with carboplatin. Three years after commencing single-agent carboplatin, he is still very well and responding to the same agent and continuous ADT (Fig. 4).

We analysed germline and somatic cancer DNA with next-generation sequencing (NGS) using a targeted panel of DNA repair genes at more than 5000-fold depth, as well as with the BROCA panel (which also covers intronic and flanking regions of genes documented to predispose to cancer) at the University of Washington (Dr. Colin Pritchard), and finally whole exome sequencing (WES) of his liver biopsy at two separate laboratories (Ann Arbor, Michigan and The ICR, London). None of these assays were able to identify pathogenic defects in canonical DNA repair genes (Fig. 5). Immunohistochemistry showed no mismatch repair deficiency or loss of ataxia telangiectasia mutation (ATM) expression. WES data analysis, however, using multiple bioinformatic approaches, confirmed a mutational signature and genome scars associated with HRDs (Fig. 6) [6]. Since ATM loss does not cause an HRD signature, these data indicate a different genomic alteration [7].

Patient 2 was diagnosed in 2002 at the age of 37 yr with adenocarcinoma of the prostate, with a PSA at presentation of 293 ng/ml, a Gleason score of 8 in prostate biopsies, and metastatic bone disease. He was commenced on ADT and treated on a clinical trial regimen at Memorial Sloan Kettering Cancer Center with six cycles of carboplatin, paclitaxel, and estramustine, achieving a PSA nadir of 0.5 ng/ml. Following a rise in his PSA to 0.7 ng/ml 6 mo later, he received radical RT to the prostate and continued ADT until 7 yr later, when he developed castration-resistant disease. Subsequently, abiraterone was initiated for 2.5 yr with a good biochemical response and radiological stability. At this time point, he was referred to the Royal Marsden; as he had a strong family history of malignancy (Fig. 7), we performed NGS of his tumour tissue from a fresh bone biopsy. This demonstrated a BRCA2 frameshift truncating mutation (c.4876_4877delAA, p.Asn1626Serfs*12), confirmed to be of germline origin, with loss of the wild type allele. He entered a phase I trial of olaparib at 200 mg bid in combination with an ataxia telangiectasia and RAD3-related protein inhibitor but had no PSA decline; however, he did have radiological stability lasting for 4 mo. In light of the extraordinary previous response to carboplatin and paclitaxel, the patient was treated again with six cycles of carboplatin (AUC 5) and docetaxel (75 mg/m²), achieving a 44% PSA decline. On diffusion-weighted magnetic resonance imaging, the majority of his metastatic bone disease showed features consistent with disease regression (including disease at T6, T7, and T12); however, a large bone metastasis replacing his entire seventh right rib presented residual hypercellular disease activity. This seventh right rib became symptomatic with pain despite having been previously irradiated with a single fraction of 8 Gy (Fig. 8). He was, therefore, advised to pursue palliative metastatic seventh right rib resection by thoracotomy. In 2018, 15 yr after his original diagnosis, the patient remains well with on-going continuous ADT and oligorecurrent disease at T6 which has recently been treated with CyberKnife irradiation.

Patient 3 was aged 66 yr in June 2015 when he was found to have a slightly elevated PSA (5.38 ng/ml) and an enlarged prostate, with biopsies confirming the diagnosis of prostate adenocarcinoma with neuroendocrine differentiation and a Gleason score of 4 + 5 (Fig. 9). A choline positron emission tomography-computed tomography scan showed positive pelvic lymph nodes and multiple bone metastases. He was treated with ADT and six cycles of docetaxel but progressed on this after 6 mo despite achieving castrate levels of testosterone. At this stage, in consideration of his strong family history (Fig. 10), he underwent genetic testing which identified a pathogenic germline mutation in ATM (heterozygous, c.2T>C [p.M1T]). Lack of ATM expression in his tumour was confirmed by immunohistochemistry. Furthermore, NGS analysis detected a TSC2 mutation (splice site donor Ex39), with loss of the other copy of TSC2, which is highly likely to be pathogenic. He was ineligible for a PARP-inhibitor trial since he developed disseminated intravascular coagulation with G3 thrombocytopenia as well as cauda equina syndrome requiring external beam radiotherapy to T12 to L3. He was subsequently commenced on single-agent carboplatin at the initial target dose of AUC 4, which was then escalated to AUC 5 after two cycles as his platelets count normalised. His PSA significantly declined from 159 ng/ml to a nadir of 36 ng/ml but then increased despite two further cycles of treatment; radiologically, there was no evidence of progression (Fig. 11). We continued treatment with carboplatin but added cabazitaxel at a dose of 20 mg/m² with a further PSA decline from 82 ng/ml to 7.6 ng/ml. He continues on the same treatment combined with continuous ADT with responding disease as of May 2018.

2. Discussion

Platinum-based chemotherapy has demonstrated impressive anti-tumour activity in some mCRPC patients; however, little published data are available on the different tumour DNA repair defects associated with platinum sensitivity. These three patients shared some atypical clinical features: a strong family history for cancer, presentation with metastatic disease, young age at diagnosis, and low PSA levels relatively to the extent of their metastatic disease. Interestingly, the first two patients were given carboplatin based solely on these clinical characteristics. The availability of NGS subsequently allowed us to elucidate the underlying genomic characteristics of these tumours.

The first patient had exhausted all of the standard treatment options and had rapidly progressive symptomatic disease before initiation of carboplatin which has given him further more than 3 yr of disease control and excellent quality of life. We were unable to identify a specific genetic defect responsible for his remarkable platinum sensitivity although WES demonstrated a mutational signature of HRD. The second patient has a germline BRCA2 mutation previously reported to confer a poorer prognosis [8]; however, impressively, he is still alive 15 yr later which compares favourably to the median survival of 6 yr on the CHAARTED trial [9] and 6.4 yr on the STAMPEDE trial [10]. These data suggest that as with ovarian cancer, platinum treatment may convert this aggressive disease subtype into a better prognosis disease. Our third patient presented with a deleterious germline ATM mutation, and his response to carboplatin arguably supports the functional impact of this genomic aberration.

In conclusion, these case series suggest the efficacy of carboplatin in metastatic prostate cancer with DNA repair defects. Further clinical trials are warranted to support its use in this subset of patients. Our report indicates that treatment selection can be facilitated by family history records as well as NGS of germline and somatic cancer DNA.



Conflicts of interest: The authors have nothing to disclose.



Funding support: We would like to acknowledge support from Prostate Cancer UK and the Movember Foundation to the London Movember Prostate Cancer Centre of Excellence at The Institute of Cancer Research and Royal Marsden (grant number CEO013-2-002), The International PCF Dream Team, Experimental Cancer Medical Centre (ECMC) grant from Cancer Research UK and the Department of Health (Ref: C12540/A25128) and Cancer Research UK (Centre Programme Grant).

Associate Editor: Giacomo Novara