Main

Insulin-like growth factors (IGFs) bind to type 1 IGF receptors (IGF-1Rs) on the surface of most cells, activating signalling effectors, including AKT, to promote normal embryonic and post-natal development (Baker et al, 1993). In tumours, IGF-1R signalling mediates proliferation, invasion and cell survival via protection from apoptosis (Chitnis et al, 2008). Activated IGF-1Rs undergo internalisation, and are then degraded or recycled back to the plasma membrane (Goh and Sorkin, 2013). We and others reported that some activated internalised IGF-1Rs translocate to the nucleus of human tumour cells, influencing gene expression (Aleksic et al, 2010; Sehat et al, 2010; Sarfstein et al, 2012; Warsito et al, 2012).

The IGF axis is recognised as mediating resistance to anticancer therapies (King et al, 2014). The first report linking IGF-1R with radioresistance came from a study of breast cancer patients, where IGF-1R overexpression was associated with early recurrence within the irradiated site (Turner et al, 1997). High IGF-1R also associates with radioresistance in cervical cancer (Lloret et al, 2007), and we reported that IGF-1R is overexpressed in HPV-negative head and neck cancers that are characterised by resistance to chemo-radiation (Dale et al, 2015). Preclinical data support a link with radioresistance: we and others showed that IGF-1R targeting enhances chemo- and radiosensitivity in a range of tumour models in vitro and in vivo, independent of apoptosis induction (Macaulay et al, 2001; Tezuka et al, 2001; Cosaceanu et al, 2007; Isebaert et al, 2011; Riesterer et al, 2011). Subsequently, we reported that IGF-1R depletion or inhibition enhances the sensitivity of prostate cancer cells to ionising radiation (IR). This effect was associated with delayed resolution of IR-induced DNA double-strand breaks (DSBs) and inhibition of DSB repair by both non-homologous end-joining and homologous recombination (Turney et al, 2012; Chitnis et al, 2014). The aim of the current study was to investigate whether these findings have clinical relevance, by assessing IGF-1R expression in prostate cancers of men treated with radical radiotherapy. Our major findings are first that patients whose tumours express high total IGF-1R experience adverse outcomes, including increased risk of overall and metastatic relapse. Second, high internalised IGF-1R associates significantly with increased risk of biochemical relapse. This group includes patients experiencing recurrence within the irradiated field, supporting a link with clinical radioresistance.

Materials and methods

Clinical cases

Clinical data were extracted from records of patients treated with external beam radiotherapy in Oxford from 2000 to 2005, including age, date of diagnosis, presenting PSA, clinical tumour (cT) stage, Gleason grade and Gleason Sum score. Using follow-up data, including serial PSA monitoring, isotope bone scans, CT scans, MRI and PET-CT scans, patients were assigned to one of three groups: long-term remission, metastatic relapse, or biochemical relapse. The latter was defined using the ASTRO-Phoenix Consensus criteria (Roach et al, 2006) as +2 ng ml−1 PSA rise above the post-radiotherapy nadir, without evidence of metastatic disease. This study was approved by the National Research Ethics Service (07/H0606/120).

Type 1 IGF receptor immunohistochemistry

Archival diagnostic formalin-fixed paraffin-embedded (FFPE) prostate biopsies were selected for this study. Routinely, left- and right-sided biopsies had been embedded in separate blocks, ⩽6 biopsies per block. For patients with unilateral tumour, the ipsilateral (i.e., cancer-containing) block was selected for study. For cases with bilateral tumour, the block selected was the most representative of Gleason grade. The chosen tumour-containing blocks were used for IGF-1R immunohistochemistry (IHC) as described (Aleksic et al, 2016). For details of method and scoring see Supplementary Information. In brief, IGF-1R signal in ⩽6 biopsies per case was scored blinded by uro-pathologist CV for intensity (0–3) and percentage positivity (0–4) in the malignant prostatic epithelium. This generated immunoreactive scores (intensity × percentage, range 0–12) for membrane, cytoplasmic and nuclear IGF-1R. Internalised IGF-1R (0–24) was represented by the sum of cytoplasmic and nuclear IGF-1R, and total IGF-1R (0–36) as the sum of membrane, cytoplasmic and nuclear IGF-1R. Where present, benign prostatic epithelium was also scored for IGF-1R content.

Statistical analysis

Clinical data were analysed using t-tests and Wilcoxon matched-pairs signed rank test for non-parametric data. The IGF-1R IHC scores were binarised by the median score and analysed as categorical variables by χ2-test. Overall survival (OS) was defined as the follow-up time from radiotherapy until death. Patients not reported as dead were censored at their last known date alive. Recurrence-free survival (RFS) was defined as time from radiotherapy until any recurrence (biochemical or metastatic). Recurrence-free survival data are displayed using Kaplan–Meier curves, with the associated log-rank (Mantel–Cox) test. Cox proportional hazards models were fitted for IGF-1R for RFS and OS in both univariate and multivariate analyses. IGF1R expression data were extracted from (Taylor et al, 2010) and analysed for association with risk of biochemical recurrence by Wald test. Statistical analysis was performed using GraphPad Prism v6, STATA 11.2 (Stata Corporation, College Station, TX, USA) and R Statistical programming environment (v3.2.4, R package: survival v2.38-3. R Core Team: Vienna, Austria). All statistical tests were two-sided, and P<0.05 was considered significant.

Results

From a database of ∼800 prostate cancer patients treated with radical radiotherapy from 2000 to 2005, we identified 136 with available FFPE tissue (Table 1), and in whom we could ascertain outcomes of remission (75 cases), metastatic recurrence (17) or biochemical recurrence (44) from imaging and serial PSAs. External beam radiotherapy had been 3D conformal CT planned, and treatment typically involved delivering a 55 Gy dose to the planning target volume in 20 fractions over 4 weeks (Table 1). Assuming an α/β ratio for prostate cancer of 1.8 Gy (Dearnaley et al, 2016), this dose/fractionation schedule is equivalent to 65.9 Gy in 2 Gy fractions. Radiotherapy was administered to all other patients in fractions of 2 Gy. Of the 136 patients, 57 (42%) had also received adjuvant endocrine therapy, generally for 6–48 months (Table 1). Median follow-up was 7.86 years (range 0.46–12.68 years). To assess the extent to which this cohort is representative, we assessed associations of adverse outcome with the principal prognostic factors: primary Gleason grade, stage, and presenting PSA (Heidenreich et al, 2014). Univariate analysis confirmed significant associations between risk of overall recurrence and Gleason grade (primary grade 4 vs 3), stage (1 vs 2 or 3) and PSA 10–20 vs <10, although increased risk in patients with PSA ⩾20 did not reach significance (Supplementary Table S1). Adjuvant endocrine therapy has been shown to improve disease-free survival and OS following radiotherapy for localised prostate cancer (Kumar et al, 2006; Bolla et al, 2010), but was associated with increased risk of overall recurrence in our series (Supplementary Figure S1A). This likely reflects the practice during 2000–2005 of offering endocrine therapy only to high-risk patients. Indeed, those receiving endocrine therapy had higher-grade tumours (median primary Gleason grade 4 vs 3) and a lower proportion of stage cT1 tumours (15 out of 57, 26% vs 30 out of 79, 38%) compared with patients not offered endocrine therapy.

Table 1 Demographic data
Full size table

We then evaluated the IGF-1R content of the prostate biopsies, detecting IGF-1R in 109 out of 136 (80.1%) of the cancers with variable subcellular localisation in the plasma membrane, cytoplasm and nuclei (Table 2; Figure 1A). All but 8 biopsies contained benign prostatic epithelium, of which 90 out of 128 (70.3%) contained detectable IGF-1R. Compared with benign glands from the same biopsies, IGF-1R expression was higher in the malignant prostatic epithelium (P<0.001; Figure 1B), confirming our previous report of IGF-1R overexpression in primary prostate cancer (Hellawell et al, 2002). The expression of IGF-1R did not associate with cT stage or presenting PSA, but higher-grade tumours (primary Gleason grade 4–5) contained significantly more IGF-1R than lower-grade tumours (primary Gleason grade 3, P=0.004; Table 2).

Table 2 Association of IGF-1R with clinical factors
Full size table
Figure 1
figure 1

Type 1 IGF receptor associates with adverse outcomes in patients treated with radical radiotherapy for prostate cancer. (A) Examples of IGF-1R IHC in diagnostic biopsies of prostate cancer from patients treated with radiotherapy. (a) Gleason 3+4=7 cancer, IGF-1R score membrane 0, cytoplasm 12, nuclear 9; (b) Gleason 4+3=7 cancer, IGF-1R score membrane 6, cytoplasm 3, nuclear 0; (c) Gleason 3+3=6 cancer containing no detectable IGF-1R; (d) Gleason 4+5=9 cancer, IGF-1R score membrane 0, cytoplasm 12, nuclear 0. Scale bar, 20 μm. (B) Total IGF-1R score (bars, mean±s.e.m.) in benign and malignant areas (***P=0.001 by Wilcoxon matched-pairs signed rank test). (C–E) Kaplan–Meier plots with log-rank (Mantel–Cox) tests to assess association of total, cytoplasmic (Cyto) and internalised (Intern) IGF-1R with (C) all recurrences, (D) metastatic recurrence and (E) biochemical recurrence.

Full size image

Using Kaplan–Meier and univariate analyses, we tested for associations between clinical outcomes and IGF-1R expression and subcellular localisation. The risk of overall recurrence was significantly greater in men whose prostate cancers contained high total IGF-1R. There were also significant associations between overall recurrence and cytoplasmic or internalised (nuclear plus cytoplasmic) IGF-1R (Table 3; Figure 1C). Analysis of membrane and nuclear IGF-1R separately showed no significant association with outcome (Table 3). Of the 136 patients, 17 (12.5%) developed scan-confirmed metastatic disease, which associated significantly with high total IGF-1R (Table 3; Figure 1D). We also assessed associations with OS, as 26 out of 136 (19%) of the cohort had died, finding borderline association between risk of death and high total IGF-1R (P=0.059; Table 3). Neither cytoplasmic nor internalised IGF-1Rs were associated with metastatic disease or OS (Table 3; Supplementary Figure 1B).

Table 3 Univariate analysis
Full size table

Biochemical recurrence was experienced by 44 (32%) patients, comparable with reported biochemical failure rates of 26–40% for low/intermediate risk disease and up to 70% for high-risk disease following radical external beam radiotherapy (Grossfeld et al, 2002; Morgan et al, 2007; Gabriele et al, 2016; Zargar et al, 2017). Biochemical recurrence showed borderline association with total IGF-1R (P=0.059 by univariate analysis, Table 3; P=0.055 by Kaplan–Meier analysis, Supplementary Figure S1C). Analysing IGF-1R scores by subcellular compartment (Table 3; Figure 1E), we identified significant associations between biochemical recurrence and IGF-1R in the cytoplasmic (P=0.002 by univariate analysis and 0.001 by Kaplan–Meier) or internalised (nuclear plus cytoplasmic; P=0.034 and 0.031, respectively) compartments. Patients experiencing biochemical recurrence include those with occult metastases and patients experiencing local recurrence within the radiation field. It is not possible to differentiate these two outcomes with certainty, because routine prostate biopsies are not performed in this situation (Morgan et al, 2007), and none of these patients underwent salvage prostatectomy. However, it is estimated that ⩾30% of post-irradiation biochemical recurrences are due to local recurrence, which is an indicator of clinical radioresistance (Bolla et al, 2010). In an attempt to further test associations of IGF-1R with local recurrence, we repeated the Kaplan–Meier analyses after excluding patients (n=5) who experienced scan-confirmed metastatic recurrence within 2 years, and were therefore likely to have had occult metastases at the time of radiotherapy. This had no effect on the significance of associations identified between risk of biochemical recurrence and high cytoplasmic (P=0.001, unchanged), internalised (P=0.030 with 5 cases excluded and P=0.031 for all 136 cases) or total IGF-1R (P=0.055, unchanged; Supplementary Figure 1D).

Using multivariate analysis, tumour stage was independently associated with overall and biochemical recurrence (Table 4), as is predictable (Heidenreich et al, 2014), although this effect was apparent only when comparing clinical stage cT2 (not cT3) with cT1. Importantly, cytoplasmic, internalised and total IGF-1Rs were independent predictors of overall recurrence. Furthermore, cytoplasmic IGF-1R and stage (cT1 vs 2) were the only factors independently associated with the risk of biochemical recurrence (Table 4; Supplementary Table S2). In this cohort, primary Gleason grade (3 vs ⩾4) was not independently associated with overall or biochemical recurrence, but was the only factor independently predictive of metastatic recurrence (Table 4; Supplementary Table S2).

Table 4 Multivariate analysis
Full size table

Finally, we utilised publically available data to test for associations between IGF-1R and outcome in men with prostate cancer treated by radical prostatectomy. Excluding 37 patients who had received previous radiotherapy, the cohort of Taylor et al (2010) included 103 patients with microarray data on IGF1R expression, and clinical data on biochemical recurrence. In this surgical data set there was no significant association between IGF1R expression and biochemical recurrence (Supplementary Figure S1E). However, as noted previously, we found only borderline association of biochemical recurrence and total IGF-1R in the radiotherapy cohort (Table 3; Supplementary Figure S1C), and we are comparing semi-quantitative IHC scores with microarray analysis of IGF1R mRNA. Therefore, it will be important to assess the prognostic significance of IGF-1R at the protein level in a surgically treated cohort, to enable a more robust comparison.

Discussion

The major findings of our study are the identification of significant associations between total IGF-1R and risk of overall and metastatic recurrence, and between cytoplasmic or internalised IGF-1R and biochemical recurrence, following radical radiotherapy but not surgery for prostate cancer. These associations were revealed by univariate and Kaplan–Meier analyses; multivariate analysis also indicated that cytoplasmic IGF-1R is an independent predictor of biochemical recurrence. Most of the patients in our series were treated pre-2006 and received radiotherapy at 55 Gy in 20 fractions, equivalent to 65.9 Gy in 2 Gy fractions. Although this seems a modest dose by current standards, this dose/fractionation schedule was widely used in the United Kingdom during the era in which these patients were treated. It remains to be seen whether IGF-1R retains predictive significance in patients treated at 74 Gy in 37 fractions as is current practice, or the biologically equivalent hypo-fractionated dose of 60 Gy in 3 Gy fractions (Dearnaley et al, 2014, 2016). Forty-two percent of patients in our series had received adjuvant endocrine therapy. This is relevant, given known crosstalk between IGF-1R, androgen receptor signalling and the DNA damage response, and reports that anti-androgen therapy enhances prostate cancer radiosensitivity in preclinical models and clinically (Pandini et al, 2005; Kumar et al, 2006; Bolla et al, 2010; Goodwin et al, 2013; Polkinghorn and Zelefsky, 2013; Zelefsky et al, 2013). In the current study, we found no evidence of better outcomes in the group receiving endocrine therapy, likely because this had been offered only to high-risk patients. It is possible that the use of adjuvant endocrine therapy in a subset of patients could have influenced the associations we found between IGF-1R and clinical outcomes.

In probing the relevance of IGF-1R subcellular localisation, we noted that the association of cytoplasmic or internalised IGF-1R with biochemical recurrence was stronger than for total IGF-1R. Type 1 IGF receptor internalisation is required for prolonged AKT activation (Romanelli et al, 2007), and for IGF-1R nuclear translocation, which contributes to IGF axis activity by promoting gene expression (Aleksic et al, 2010; Sehat et al, 2010; Sarfstein et al, 2012; Warsito et al, 2012). The lack of significant association with nuclear IGF-1R alone, and the stronger association of cytoplasmic IGF-1R than internalised (nuclear plus cytoplasmic) IGF-1R with overall and biochemical recurrence (Table 3) suggest that cytoplasmic IGF-1R is the major contributor here. Supporting this view, cytoplasmic IGF-1R was an independent predictor of biochemical recurrence in multivariate analysis (Table 4). Therefore, it is plausible that the association of biochemical recurrence with cytoplasmic IGF-1R could reflect increased IGF-1R activation in radioresistant tumours. This would be consistent with data we reported in cell lines, where suppression of IGF-1R activity enhanced radiosensitivity (Turney et al, 2012; Chitnis et al, 2014). In patients with PSA-detected localised prostate cancer, radiotherapy and surgery are reported to lead to comparable outcomes at median 10 years’ follow-up (Hamdy et al, 2016). Several predictive biomarkers have been identified to predict outcomes post irradiation for prostate cancer (Wilkins et al, 2015), but a role for IGF-1R has not previously been highlighted in this context. Our data suggest that patients whose prostate cancers contain high cytoplasmic or high total IGF-1R should be offered surgery or dose-escalated radiotherapy, hypotheses that warrant prospective evaluation. Taken together with preclinical data (Isebaert et al, 2011; Turney et al, 2012; Chitnis et al, 2014), these results also support evaluation of IGF axis inhibition as a novel route to radiosensitisation of prostate cancers that express high total or cytoplasmic IGF-1R.