Serum Insulin-Like Growth Factor-I Is Positively Associated with Serum Prostate-Specific Antigen in Middle-Aged Men without Evidence of Prostate Cancer

Serum insulin-like growth factor-I (IGF-I), a powerful mitogen, is positively associated with an individual’s risk of developing breast and prostate cancer (1). Higher circulating levels of IGF-I have also been related to greater mammographic density in women without breast cancer, and it has been postulated that IGF-I may increase breast cancer risk in part by stimulating epithelial proliferation (2). We have investigated whether IGF-I is similarly associated with cellular proliferation in the prostate gland. In men with no evidence of prostate cancer, levels of prostate-specific antigen (PSA) correlate closely with prostate size and the presence of benign prostatic hyperplasia (BPH; Ref. 3). Although BPH, unlike mammographic density, is not clearly linked to cancer risk, it is a condition typified by increased proliferation of epithelial and stromal tissue. We, therefore, examined the association between serum IGF-I and PSA among a group of 367 healthy men without evidence of prostate cancer.

Participants had been selected as controls in a case-control study exploring risk factors for localized prostate cancer (4). All of the men had serum PSA measured as part of a case-detection exercise for a randomized controlled trial of treatment of localized prostate cancer conducted in three United Kingdom centers, the ProtecT study (5). Men with a normal PSA (<3 ng/ml) or with a PSA ≥3 ng/ml and at least one negative prostatic biopsy were eligible as controls, and were age-matched to cases. The relationships between IGF-I, PSA, and age were examined using least-squares linear regression and by calculating Pearson’s correlation coefficients. The distribution of PSA was strongly positively skewed and log PSA was, therefore, used in analyses. Because the age distribution of participants was conditional on that of cases, rather than the general population, weights were applied to regression analyses reflecting the inverse probability of sampling by age group. All of the regression analyses were adjusted for age and study center.

The mean level of IGF-I for each quartile of the PSA distribution, adjusted for age and study center, was derived from linear regression. The prevalence of BPH rises steeply with age, and we examined the possibility that associations between IGF-I and PSA differed in men ages <60 years and ≥60 years by fitting interaction terms to our models.

The mean age of participants was 62 (range, 50–70) years, median PSA was 1.2 ng/ml (interquartile range, 0.7–2.0 ng/ml), and median IGF-I was 123.1 (interquartile range, 97.6–149.5 ng·ml). Age was positively associated with PSA (correlation coefficient, r = 0.16; P = 0.002) and showed a weak inverse association with IGF-I (r = −0.08; P = 0.14; Figs. 1 and 2). Overall, a significant positive relationship was observed between serum IGF-1 and PSA levels (r = 0.14; P = 0.006; Figs. 3 and 4). The change in IGF-I per quartile increase in PSA was 4.3 ng/ml (95% confidence interval, 0.1–8.6; P_trend = 0.05), there was no evidence of nonlinearity. This relationship was stronger in older men (Fig. 5). In men ages 60 years and over (n = 258), the change in IGF-I per quartile increase in PSA was 7.3 ng/ml (95% confidence interval, 3.4–11.3), compared with 1.7 ng/ml (95% confidence interval, −5.3–8.6) in men ages 50–59 (n = 109). A formal test for interaction bordered on statistical significance (P_interaction = 0.05).

The positive association of PSA with IGF-I in men with no evidence of prostate cancer is consistent with IGF-I stimulating the proliferation of epithelial cells within the prostate. The induction of epithelial proliferation would increase the number of cells at risk from carcinogenic insults and thereby provide a mechanism through which IGF-I could increase prostate cancer risk (1). Our findings are also consistent with a recent study conducted in Shanghai that observed a positive relationship between higher circulating IGF-I levels and symptomatic BPH (6). The observation that the association between IGF-I and PSA was stronger in older men may have a number of explanations. In younger men, PSA may simply reflect the size of the prostate gland, whereas in older men, variability in PSA may be driven by the presence or absence of prostatic hyperplasia, and it may be epithelial hyperplasia that underlies the IGF-I/PSA association. Such an explanation is consistent with the observation of a positive association between IGF-I and mammographic density in women before, but not after, the menopause (2). It is in the premenopausal period that cellular proliferation is occurring in breast tissue; in the prostate gland, epithelial and stromal proliferation increase in many older men.

The possibility that the association with IGF-I results from PSA produced by occult cancer cannot be entirely discounted, but is less likely inasmuch as all men with a PSA ≥3 ng/ml had normal prostatic biopsies. In addition, among older men, the association between IGF-I and PSA was just as evident across ranges of PSA of less than 2 ng/ml (see Fig. 5).

IGF-I production by prostate tissue is relatively small and is unlikely to influence the large circulating pool of IGF-I, even in the presence of BPH. Although PSA is a proteinase and can, in semen, cleave the major binding protein for IGF-I, insulin-like growth factor binding protein (IGFBP)-3, it is inactivated in the circulation and is unlikely to affect serum IGF-I. Furthermore, in situations in which IGF-I is released by carrier protein cleavage, serum levels actually decline through increased clearance.

In conclusion, we have shown a positive association between IGF-I and serum PSA in healthy middle-aged men, consistent with IGF-I stimulating the proliferation of cells within the prostate. If confirmed, this association may indicate a biological pathway influencing risks of both benign and malignant prostatic disease, and provide an opportunity for risk reduction through therapeutic or lifestyle modification of the IGF-axis.

We are pleased to thank Sara Bright, Zoe Wilkins, and Tracey Calthorpe for clerical support; Mark Sidaway and Dan Dedman for database management; and all staff within the ProtecT research team for their help and cooperation.