Ex vivo γH2AX assay for tumor radiosensitivity in primary prostate cancer patients and correlation with clinical parameters

Prostate cancer (PCa) still remains the most common type of cancer diagnosed in men in the western world [1] and almost 40% of men aged older than 65 years undergo radiation therapy (RT) as a curative therapy [2]. PCa patients with primary whole gland radiation therapy (RT) have a 10–30% probability of biochemical relapse [3]. New RT concepts, such as focal dose escalation [4] or escalation of systemic therapy [5] improve the outcomes for high-risk non-metastatic PCa patients. RT management decisions are usually based on three pre-therapeutic factors: Prostate Specific Antigen (PSA) serum concentration, Gleason Score (GS) and tumor stage (T stage). They currently represent the most common prognostic factors for the patient’s outcome and are therefore routinely used in the risk stratification of PCa patients [6]. However, none of the latter risk factors was described to correlate with the intrinsic radio sensitivity of PCa lesions [7]. The intrinsic radio sensitivity plays a major role in the therapeutic response to RT and its characterization and quantification might enable individual dose prescription concepts on a lesion or even on a voxel level in the future.

Ionizing radiation (IR) targets mainly the chromosomal DNA and induces DNA breaks directly and indirectly through water radiolysis products. The most common DNA damage patterns due to RT are double-strand breaks (DSBs) [8]. One sensitive tool for measuring the amount of DNA DSBs and therefore quantifying the impact of radiotherapy is detecting the γH2AX foci, a histone which becomes rapidly phosphorylated after exposition to IR [9].

Previous studies [7, 10, 11] established a method to distinguish between radio sensitive and radio resistant PCa lesions, quantifying residual γH2AX foci in ex vivo irradiated tumor samples. Their results indicated a high inter-lesion heterogeneity for intrinsic radio response, suggesting the necessity of personalized RT methods. Following the same methodology, we collected three biopsies from PCa patients during High Dose Rate Brachytherapy (HDR-BT) procedure. In two biopsy cores, the amount of the γH2AX residual foci was determined after ex-vivo irradiation with 0 and 4 Gy for comparison of inter-lesional differences in radio sensitivity. The third core was used for genomic analyses within the PCa tissue. Lastly, the correlation between different clinical parameters, imaging parameters and gene mutations with radio response was examined.

Figure 1 shows the distribution of residual γH2AX foci for the 18 PCa lesions, 24 h after ex-vivo irradiation. The median value of the normalized γH2AX foci number for all patients was calculated (3.12, range: 0–10.74) and the patients were divided into 2 groups according to this value: radio resistant vs radio sensitive. The median normalized γH2AX foci value was 5.03 (range: 3.18–10.74) and 2.5 (range: 0–3.06) for the radio sensitive and radio resistant group, respectively (p < 0.001). Two patients showed two different bilateral tumor areas, respectively. In one patient, one lesion was perceived as radio sensitive (median normalized γH2AX foci: 5.03) whereas the contralateral biopsy core was relocated to the radio resistant group (median value: 2.64). The other patient showed no difference in intrinsic radio sensitivity between the two tumor areas (p = 0.572). Figure 2 shows the heterogeneous response to RT between the patients and lesions. To investigate the possible predictive markers for radio sensitivity and radio resistance, we proofed the association of the acquired clinical and genetic imaging factors to each group (Table 2). No significant difference between the two groups was observed regarding age, GS, previous ADT, cT stage, absolute GTV volume and ADC-max values. The PSA serum concentration as well as the SUV-max and SUV-mean values were significantly higher in the radio sensitive group. ADC-mean values were significantly smaller in the radio sensitive group. Figure 3 shows the results of the Pearson’s correlation test for the significant parameters.

The tumor DNA samples from nine patients were analyzed with NGS. In turn, the tumors of four patients possessed mutations. However, three patients carried mutations in ATM, BRCA1 and PMS2 genes that were classified as a variant of unknown significance (VUS), according to the ClinVar database. Finally, only one patient displayed a previously not reported loss of function frameshift mutation in the NBN gene (c.654_658delAAAAC) that introduces a premature termination codon and results in a truncated protein (p.Lys219AsnfsX16). The NBN c.657_661delACAAA mutation that results in the same truncated protein as the one reported herein has been reported as pathogenic in the ClinVar database, is a founder variant among Central and Eastern European populations and has been associated with breast and PCa [21, 22] (Fig. 4).

For the cT classification three patients were excluded (6 and 12, with bilateral lesions and where ADC sequence was not available). cT staging was conducted via PSMA PET and mpMRI, respectively.

In the current study, we used the γH2AX assay, established by the group of Menegakis et al. [12] to determine the intra- and inter-tumoral heterogeneity regarding intrinsic radio sensitivity in PCa patients.

Current RT methods are characterized by a homogeneous dose distribution within the prostate. However, biochemical recurrence free survival (bRFS) has been shown to be improved by focal dose escalation based up to 7% for intermediate- and high-risk prostate cancer without impacting toxicity and quality of life [5]. An individualized dose prescription according to the respective radio resistance of each tumor lesion could be a further step towards individualized RT in PCa. In our study we found significant differences in radio resistance between PCa lesions in different patients. Interestingly, also a difference in radio resistance was found within two PCa lesions of the same patient. Our results are similar with the observations by DeColle et al. [14]. Taken together the current evidence suggests the necessity of individual RT concepts, due to the inter-tumoral heterogeneity in intrinsic cellular radiation sensitivity. However, quantification of radio resistance by measurement of residual γH2AX foci may not be suitable for clinical routine due to its labor-intense nature. Likewise, clinical surrogate parameters are warranted for its prediction.

Therefore, we investigated the possible predictive characteristics of the patients’ clinical and imaging variables for radio resistance. No significant difference was observed between intrinsic radio sensitivity of PCa samples and cT stage, PCa lesion volume, ADT admission or GS, which is consistent with the results of De Colle et al. [14]. However, our results showed significantly higher PSA serum concentrations in the radio sensitive group than in the radio resistant patients and strong negative correlation in Pearson’s test. Apart from its diagnostic and prognostic features, PSA has been showed to have an important role in PCa proliferation [23] and it is well known that radiation sensitivity increases with proliferation due to telomere dysfunction [24]. Additionally, SUV-max and SUV-median values in PSMA PET images were significantly higher in the radio sensitive group, whereas ADC-median values were significantly smaller. All values showed also a significant correlation with nfoci number with r ≥ 0.60 in Pearson’s test. This suggests a higher PSMA expression and a higher cell density in radio sensitive PCa lesions. Nevertheless, first studies reported that high SUV-max values in primary PCa lesions are associated with an increased relapse rate after surgery [25]. These findings must be discussed under consideration of the biological properties of the PSMA protein [26, 27]. PSMA on PCa cells hydrolyzes poly-y-glutamated folates and increases the glutamate and folate concentrations within PCa cells [26]. Additionally, the expression of PSMA correlates with the PI3K-Akt pathway [28]. Both mechanisms are associated with an increase in tumor proliferation. Furthermore, PSMA contributes in tumor angiogenesis [29]. Likewise, it can be assumed that PCa lesions with high PSMA expression might possess a higher metastatic potential and thus a faster systemic disease progression which might explain worse outcomes after RT despite an increased radio sensitivity. After further validation, this observation may change treatment concepts in PCa patients. In patients with high SUV values in PSMA PET imaging, low ADC values in MRI and high PSA levels an escalation of systemic treatment and a reduction of RT dose might increase the therapeutic ratio.

The correlation between mutations in known PCa susceptibility genes (such as BRCA1 and 2, ATM, PMS2 and NBN) and an increased risk for PCa was showed in multiple studies [30‐32]. Moreover, mutations in the NBN, BRCA2 and ATM genes are associated with a more aggressive PCa phenotype and worse clinical outcome [21, 33, 34]. In the current study, one sample showed a significant NBN gene mutation. The PCa specimen was derived from a patient classified as radio sensitive, with a median number of foci of 3.95 and a high risk PCa, according to the clinical parameters. The encoded protein (nibrin) is a component of the MRE11/RAD50/NBS1 (MRN) complex, which is thought to be involved in DNA DSB repair and DNA damage-induced checkpoint activation [35]. In a prospective study, Berlin et al. investigated the role of NBN gene mutations on the clinical outcome of PCa patients treated with image-guided radiotherapy and radical prostatectomy [36]. The 5 years biochemical relapse free rate in irradiated patients was significantly higher in patients with NBN gain mutations compared to neutral gain mutations. However, NBN gain did not have any significant prognostic value in the surgery cohort. Therefore, one can speculate that, while NBN gain influences the intrinsic tumor radio resistance, NBN loss of function mutations may contribute to the sensitization of PCa to radiation. However, more studies need to be carried out in order to investigate the significance of the NBN gene mutation on PCa.

In the following, we want to discuss the limitations of our study. First, this pilot study considered only 18 PCa lesions. This is justified by considering the labor-intensive processing required for each patient. An enlargement of the cohort up to 50 patients is currently ongoing. Second, only one biopsy core per PCa lesion was irradiated with 4 Gy and thus intra-tumoral heterogeneity was not accounted for. However, De-Colle et al. demonstrated that one biopsy is sufficient to estimate the mean value of residual γH2AX foci per dose level and account for intra-tumoral heterogeneity [7].

Finally, the time difference between the estimation of the PSA serum concentration and the biopsy core recovery should be considered (median time difference of 69.5 days with a range of 6–205 days). Moreover, SUV-mean and SUV-max and ADC-mean values showed potential predictive features in detecting the eligibility of patients for RT. Nonetheless, the pre-therapeutic images and the acquiring of the biopsies during the first HDR-BT session were not performed with an equal time difference for all patients. The median number of days between the imaging and the first HDR-BT session was 105.5 days (range: 6–180) and 43 days (range: 4–200) for PSMA-PET and mpMRI, respectively.

In this pilot study we investigated the inter-tumoral heterogeneity in radio resistance in PCa patients and novel radiomic and genomic biomarkers correlated with the intrinsic radio resistance on a tumor level. After further prospective validation in larger patient cohorts, these findings might be used in the future for personalized RT concepts with individual RT dose prescriptions.