The Influence of Exercise on the Insulin-like Growth Factor Axis in Oncology: Physiological Basis, Current, and Future Perspectives

Exercise prescription characteristics

The characteristics of exercise interventions are displayed in Table 1. Three studies utilized aerobic exercise (29, 31, 34), three used resistance exercise (30, 34, 35), one study used a combination of both aerobic and resistance exercise (33), and one study used a Tai Chi Chuan exercise protocol (32). Only one trial (34) compared two types of exercise prescription on the IGF axis. Duration of the interventions ranged from 12 to 26 weeks, with the frequency of prescribed sessions ranging from 2 to 5 sessions per week. Intensity of exercise varied according to the mode; however, aerobic exercise was predominantly of moderate intensity and was quantified according to heart rate, power outputs equivalent to oxygen consumption (O₂) thresholds, and ratings of perceived exertion. Studies using resistance-training protocols (30, 34, 35) progressively altered the prescription of sets, repetitions, or intensity throughout the intervention. Resistance training prescription ranged from 2 to 4 sets at an intensity of 6 to 12 repetition maximum (RM), which was considered to be of moderate to high intensity. One intervention (33) prescribed a specific dose of weekly physical activity; however, specifications for frequency, duration, or intensity of exercise were not provided.

Effects of exercise prescription on the IGF axis

The effects of exercise interventions on IGF1 and IGF2 are displayed in Table 2, and IGFBP1, IGFBP2, and IGFBP3 in Table 3. IGF1 was measured in all studies, with decreases (29, 31, 34), increases (33), and nonsignificant changes reported (30, 32, 35). In response to similar aerobic exercise interventions in breast cancer survivors, two studies (29, 31) demonstrated a significant decrease in IGF1 of 7.3% (P = 0.045) and 2.9% (P = 0.026) from baseline, respectively. One study (34) found that 13 weeks of resistance training significantly reduced serum IGF1 by 13.1% (P ≤ 0.05) in prostate cancer patients with no further significant changes after an additional 13 weeks of training. No within-group changes in IGF1 were observed in response to the parallel aerobic training group (P > 0.05). A significant increase (17.8%; P = 0.007) in IGF1 was observed in a group of colorectal cancer survivors following a home-based exercise protocol of combined aerobic and resistance exercise (33). No other significant changes in IGF1 were observed in the remaining studies (30, 32, 35).

One study (35) reported a significant decrease in plasma IGF2 (3.0%; P = 0.02) in response to 26 weeks of progressive resistance training. A second study (29) reported similar decreases (2.4%) in IGF2 in response to aerobic exercise; however, this was not statistically significant (P = 0.101).

The most frequently measured binding protein, IGFBP3, was reported to both increase (29, 33, 34), decrease (31, 34), and remain unchanged (32, 35) in response to exercise training. A significant 4.8% increase in IGFBP3 (P = 0.021) was reported in response to an aerobic exercise intervention in breast cancer survivors (29). Conversely, in response to a similar aerobic intervention in breast cancer survivors, a significant 4.1% decrease (P = 0.006) was observed compared with the control group (31). Baseline IGFBP3 levels were substantially different among trials, ranging from 2,160.8 ng.mL⁻¹ (29) to 4,150.0 ng.mL⁻¹ (31). A significant 30.3% within-group increase (P = 0.013) in response to a home-based resistance and aerobic exercise program was observed in colorectal cancer survivors (33). In the group of prostate cancer patients completing 6 months of either home-based aerobic or resistance exercise program, a significant 12.1% (P ≤ 0.05) within-group increase in serum IGFBP3 was observed in response to resistance training and a 23.7% (P ≤ 0.05) within-group decrease in response to aerobic training (34). No significant changes in IGFBP3 were reported in the remaining trials (32, 35). Reported IGFBP3 values in one study (32) were approximately 100-fold lower than all other trials, the reason for which is not clear.

No significant changes in IGFBP1 or IGFBP2 were reported in response to aerobic, resistance, or Tai Chi Chuan–based exercise in breast cancer survivors (29, 32, 35). In addition no significant (P = 0.35) changes in IGFBP1 were observed in colorectal cancer survivors (33).

With only seven articles meeting the inclusion criteria and the heterogeneity of results, there is currently insufficient evidence to categorically determine the effects of exercise, and, more specifically, exercise prescription on the IGF axis in populations with a history of cancer. Results from two trials conducted in breast cancer survivors showed a decrease in IGF1 in response to exercise compared with the control groups (29, 31). Beyond these decreases however, the IGF1 changes in other studies changes were remarkably heterogeneous with substantial variation in the magnitude and direction. This heterogeneity of findings extended to changes in IGFBP3 with bidirectional responses reported following exercise interventions. Notwithstanding differences among intervention populations and exercise interventions, the current scope of literature does not allow for discrete conclusions to be made regarding the effects of exercise on systemic IGF1 and IGFBP3. Importantly however, these findings do not preclude a possible relationship, rather the disparity strongly indicates that the relationship between the IGF axis and exercise may be nonlinear and influenced by numerous factors, the complexities of which are discussed below.

Insights from nononcological populations

Although IGF axis changes have been assessed in populations with a history of cancer, several other studies have investigated these responses in healthy populations. Studies of healthy populations have reported similar inconsistencies in IGF axis response to exercise to those found in the oncology literature (36, 37). No changes in IGF1 or IGFBP3 were observed in similar year-long aerobic exercise interventions by McTiernan and colleagues (ref. 38; 45 minutes; moderate intensity; 5 days/week) and Friedenreich and colleagues (ref. 39; 45 minutes; 70%–80% heart rate reserve; 5 days/week) compared with controls. Utilizing a shorter-duration intervention, Arikawa and colleagues (40) demonstrated a small increase in IGFBP3 in response to an aerobic exercise intervention [30 minutes; 80%–85% maximum heart rate (HR_max); 5 days/week; 16 weeks] in young women, with no concurrent change in IGF1, IGFBP1, or IGFBP2. In contrast, decreases in systemic IGF1 levels have been reported in response to high volume aerobic training (60 minutes; cycling at lactate threshold; 5 days/week; 6 weeks; ref. 41). Collectively, studies investigating the effect of exercise training on systemic IGF axis levels in both healthy populations and populations with a history of cancer have demonstrated inconsistent or nonsignificant findings.

Mechanistic determinants of the exercise response

This review demonstrates a paucity of data and a remarkable heterogeneity when considering the response of the IGF axis to exercise. An important caveat and possible determining factor to explain the discrepancies in the IGF axis response to exercise is baseline concentrations of the IGF ligands. Orsatti and colleagues (42) and Nishida and colleagues (41) investigated the effects of resistance and aerobic training protocols, respectively, on the IGF axis; however in contrast with previous investigations, they provided exploratory analyses into the individual responses of IGF1. Exercise-induced changes in systemic IGF1 from baseline to endpoint (delta percentage values) were linearly correlated (r = −0.62, ref. 42; r = −0.77, ref. 41) with resting preintervention IGF1 levels, demonstrating that participants with elevated baseline IGF1 experienced the greatest decrease in response to exercise, whereas those with lower baseline levels of IGF1 experienced increases in response to training. These findings present an intriguing hypothesis for the specific role of exercise in the optimization of the systemic IGF axis and subsequently provide greater scope for the interpretation of the results of the current review. Exercise may facilitate increases in systemic IGF ligand availability during deficient states to an optimal level at which the anabolic potential of the systemic IGF axis is maximized (43). After this optimization, the autocrine intramuscular IGF axis may become more important in facilitating exercise-induced adaptation with minimal systemic changes (44–46). Conversely, in states of IGF ligand overabundance, exercise may act to reduce systemic IGF1 levels. This holds particular importance when considering that it is the deregulation and overexpression of IGF1 and IGF2 that is associated with the development and progression of carcinomas. Decreasing IGF concentrations in states of systemic overabundance may effectively reduce IGF1R–mediated signaling that can subsequently improve cancer outcomes through ablation of this mitotic signaling pathway. With this physiologically plausible dichotomy of the IGF ligand responses in mind, it seems appropriate that in populations with a relatively normative expression of these proteins, it may be expected that there would be no exercise-mediated changes, an idea briefly suggested by Schmitz and colleagues (35). Comparatively, in populations with either elevated or reduced expression of IGF ligands, responses to exercise may be substantially more dynamic.

A key determinant of this hypothesis regarding changes in IGF ligand concentrations in response to exercise is what constitutes “abnormal” levels of the IGF ligands, which is predicated on understanding the factors that regulate the expression of these proteins. A principle determinant of both resting IGF1 and IGFBP3 levels is age. Both IGF1 and IGFBP3 demonstrate a continual increase to a pubertal peak before a gradual decline with aging (47, 48). Accounting for the population age is therefore critical to understanding the normative systemic expression of both IGF1 and IGFBP3 and when comparing between populations. In the current literature, there was a significant range in the age of populations, with mean age ranging from 53 to 72 years. Future trials should endeavor to utilize populations of a homogeneous age to reduce variability in the IGF axis related to age, or comparing against age-standardized reference ranges. However, caution should be extended when comparing populations with previously published reference ranges as these are often dependent on the assay method used for analysis, with considerable interassay variation reported (49). In recognition of this interassay variation, an international consensus statement on the standardization of IGF1 quantification was developed, which recommended a common IGF1 standard (02/254) for assay calibration (50). Future trials should ensure that measurements of IGF or any comparisons with reference ranges adhere to these recommendations to ensure accuracy and comparability of results. In the present review, all studies except one (32) used an ELISA method for IGF ligand and binding protein measurement. Reference ranges for measures of IGF1 and IGFBP3 derived from the ELISA method adhering to the consensus statement are currently only available for pediatric populations (51). One set of IGF1 reference ranges published prior to this consensus statement by Andreassen and colleagues (52) utilizing the ELISA method reported near-identical standard curves when comparing the manufacturer-provided standard with the 02/254 standard and offers an acceptable reference range for IGF1 levels until more extensive data are published. When considering the IGF1 levels of studies included in this review against this reference range, the decrease in IGF1 reported in the trial by Irwin and colleagues (31) was from a baseline value 109.1% greater than the mean of the age- and gender-matched range. The values reported by Santa Mina and colleagues (34) were approximately 85% higher than the reference mean; however, a significant decrease was only reported from baseline to midpoint in response to resistance training. Interestingly, the decrease reported by Fairey and colleagues (29) was from an initial value approximately 33.2% below the mean for this age group, suggesting that factors beyond baseline levels likely influenced the IGF1 response.

Interpreting changes in IGFBP3 with reference to the baseline values allow for some important observations. Specifically, increases in IGFBP3 tended to be reported in groups with baseline values between 2,000 and 3,000 ng.mL⁻¹ (29, 33), whereas those that reported decreases had baseline levels of greater than 4,000 ng.mL⁻¹ (31, 34). The influence of baseline values appears particularly evident in the results reported by Santa Mina and colleagues (34), who demonstrated an IGFBP3 increase in response to resistance training and a decrease in response to aerobic training. Before concluding that differential effects exist between exercise modalities, is should be noted that baseline IGFBP3 values in the aerobic group were significantly greater (28%) than those in the resistance-training group. This disparity may explain, at least in part, why these opposing changes were observed. In the absence of extensive reference ranges utilizing the ELISA method, this method of interpretation should only be considered exploratory until further data are available.

IGF1 levels have been positively correlated with measures of body mass index, and although at the population level, this effect was not of sufficient clinical significance to warrant weight-standardized reference ranges (47), at the individual level, this may be important when considering the IGF axis response to exercise. Synthesis of IGF1 occurs in a range of extrahepatic tissues, such as skeletal muscle, bone, cartilage, and adipose tissue (53, 54). Although the main contributor to systemic levels is hepatic production, changes in the volume of tissue producing IGF1 in an autocrine fashion may also translate to systemic alterations. With exercise, systemic IGF1 alterations resulting from autocrine production are primarily concerned with changes in tissues that are also exercise-responsive, such as skeletal myocytes and adipocytes. Increases in skeletal muscle IGF1R density and activation in response to exercise should also be considered as a potential mechanism for reducing systemic levels of IGF1. Skeletal muscle IGF1R density and activation should be interpreted concomitantly with measures of muscle volume, as the surface area of IGF1–permeable myocytes will strongly influence the relationship between exercise and systemic IGF ligand alterations. Although body fat (fat mass or body fat percentage) and muscle volume were reported in 71.4% and 42.9% of studies, respectively, only one study (35) utilized the more accurate dual-energy x-ray absorptiometry method as compared with bioelectric impedance or skinfold measures (55). Given the potential influence of body composition on the relationship between the IGF axis and exercise, quantification of concurrent changes in these body composition parameters beyond basic anthropometric measures to include more accurate and reproducible measurements is recommended for future exercise interventions measuring the IGF axis.

Importantly, the potential nonlinearity of the IGF axis response to exercise and its relationship to age and body mass may necessitate the inclusion of more individualized analytics in preference to strictly population- or group-based analysis. In several studies (29, 31–33, 35), there was substantial variance both within groups at baseline and in the magnitude of change in response to exercise for several IGF axis markers. Specifically, the baseline IGF1 coefficients of variation in the exercise groups by Fairey and colleagues (29) and Lee and colleagues (33) were 43.2% and 44.5%, respectively. Furthermore, coefficients of variation of the mean changes in systemic IGF1 levels in response to the interventions ranged from 75.9% to 218.4% in the four studies reporting these data (29, 31, 32, 35). Similar variances in the mean changes in IGFBP3 in response to exercise were also reported (42.1%–519.3%; refs. 29, 31, 35). Collectively, these data suggest considerable individual IGF axis variation at baseline and in response to exercise, with important determinants of this response likely overlooked by population analysis. In addition to age and body mass, differences in time since diagnosis and treatment, as well as types of treatment may contribute to the observed variability. Mean time since diagnosis was only reported in three studies and ranged from 1.5 to 3.6 years (31, 33, 35). Average time since treatment was only reported in two studies (29, 35), ranging from 1.1 to 1.2 years. The majority of studies reported information relating to stage of disease (29, 31, 33–35), with some also explicitly specifying a range of stages (most commonly stages I to III) as part of the inclusion criteria (29, 31–33). Irwin and colleagues (31) specifically analyzed the effect of breast cancer stage on the IGF axis response. Although this factor did not modify the effect of exercise, accounting for stage of disease may provide additional insight when investigating the effects of exercise on the IGF axis. Although participants in the included investigations were not currently undergoing surgical, chemotherapeutic, or radiation treatment, current hormone therapy use was prevalent among breast cancer survivors, with 46% (29), 81% (35), and 65% (31) of participants undergoing some form of hormone therapy reported in three of the four trials. All participants (100%) in the studies of men with prostate cancer were undergoing current androgen deprivation therapy for in situ carcinomas (30, 34). For clinically meaningful subanalyses to occur according to these potential determinants of the variance observed in the IGF axis response to exercise, studies need to be of a sufficient sample size. In the present review, no studies reported sample size or power calculations, likely due to the predominance of pilot studies; however, there was considerable variation in the sample size of the included investigations. The sample size of single-group intervention studies ranged from 10 to 17 participants (30, 33), whereas randomized controlled trials ranged from 19 to 85 participants (29, 31, 32, 34, 35). It is strongly recommended that future trials ensure that the sample sizes are sufficient to allow for subanalyses to be performed, providing greater scope for understanding the relationship between the IGF axis substituents and various moderating factors. Given our developing understanding of the IGF axis response to exercise, the inconsistency and heterogeneity of current data make it difficult to perform power calculations for informing sample sizes. In lieu of this, we recommend that study populations be drawn from a sample that minimizes the variability associated with possible influencing factors of the IGF axis response, particularly stage of disease, time since diagnosis, and treatment as well as types of treatment to enhance the homogeneity of the population. We also recommend that these data be more consistently reported. Inclusion of meaningful subanalyses in future well-powered trials may enable detection of more subtle relationships between exercise and the IGF axis in this population and add greater scope for understanding the inconsistency of current results.

Differential physiologic roles of the IGF axis in response to exercise

A limitation of oncological research investigating the IGF axis in response to exercise is the exclusive focus on carcinogenic properties of the systemic IGF ligands (54, 56). Within the scope of characterizing the effect of exercise on the IGF axis, it is critical to recognize the noncarcinogenic, anabolic roles of the IGF axis in exercise-responsive tissues within the body. This is particularly important as the extrapolation of this understanding is integral to the physiologic basis by which exercise may influence the systemic IGF axis. Similar to the carcinogenic potential of the IGF axis mediated through IGF1R activation, the anabolic potential of the IGF axis is ascribed to the activation of this receptor in myocytes to promote cellular proliferation and growth (57). A point of contention is the relative contribution of systemic as compared with autocrine IGF1 in initiating these anabolic processes and how these processes are influenced by exercise. In vivo murine models have provided key insights into the isolated importance of systemic and local muscular IGF1 expression in facilitating anabolic adaptations to exercise. In a liver-specific IGF1–deficient murine model (expressing 70% reductions in systemic IGF1 compared with wild-type models), serological IGF1 deficiency did not impair improvements in muscular performance in response to exercise training (58), whereas intramuscular overexpression of local IGF1 accelerated muscular regeneration after a period of immobilization-induced atrophy in IGF1 normative mice (59). In clinical populations characterized by cachexia and muscular atrophy, local IGF1 has been shown to mediate the anabolic response of skeletal muscle to exercise (60–62). Hambrecht and colleagues (60) demonstrated local muscular IGF1 increased by 81% with no parallel change in systemic IGF1 in response to 6 months of aerobic exercise training [20 minutes/day; 70% maximal oxygen uptake (O₂max)]. Utilizing a progressive resistance-training program (3 × 8 repetitions; 80% 1RM; 2 days/week; 24 weeks), Lemmey and colleagues (61) found significant increases in total lean and appendicular muscle mass in older adults with rheumatoid arthritis. These changes were associated with 41% and 73% increases in muscular IGF1 and IGFBP3, respectively, whereas no changes in systemic IGF1, IGF2, IGFBP1, or IGFBP3 were observed.

To further explore this relationship, it must be noted that IGF1 is expressed in three specific isoforms: IGF1Ea, IGF1Eb, and IGF1Ec. IGF1Ea and IGF1Eb are splice variants typically found in systemic circulation (primarily of hepatic origin), whereas the IGF1Ec is an autocrine variant exclusively produced locally in myocytes. IGF1Ec is upregulated in response to mechanical stimuli (e.g., exercise) leading to this isoform being termed mechano-growth factor (MGF; ref. 43). In younger men (29.5 ± 1.5 years), Hameed and colleagues (63) demonstrated acute increases in MGF in response to high-resistance knee extension (10 × 6 repetitions; 80% 1RM) with no changes in IGF1Ea. Ahtiainen and colleagues (64) reported acute increases in both MGF and IGF1Ea in response to heavy-resistance exercise (5 × 10RM leg press and 4 × 10RM squats). Conversely, Popov and colleagues (65) found that only a high-intensity stimulus (75% 1RM) was sufficient to acutely increase MGF levels, with no changes observed in response to a moderate intensity [50% 1RM] and no changes in IGF1Ea reported in either condition. Although further research is required to determine the contribution of the IGF axis to anabolic adaptation, it is generally thought that improvements are mediated through local rather than systemic expression (36, 66) with MGF the primary regulator of these processes in response to exercise (43, 67, 68). Furthermore, local myocyte expression of the IGF1R has been shown to increase in response to acute resistance exercise (69, 70). Wilborn and colleagues (71) demonstrated an identical increase in the expression of muscular IGF1R mRNA in response to both higher intensity (80%–85% 1RM) and lower intensity (60%–65% 1RM) resistance training, peaking 2 hours after exercise and returning to baseline after 6 hours. In the same cohort, Taylor and colleagues (72) demonstrated elevated IGF1R phosphorylation up to 6 hours after exercise in skeletal muscle. It has been reported that during exercise, active skeletal muscle surpasses the liver as the principle producer of systemic IGF1 (autocrine production) and also is the principle extractor of IGF1 from circulation (43, 67). Increases in the expression and activation of the IGF1R may potentiate a transient postexercise period of increased IGF ligand permeability and subsequent cellular signaling. This increase in receptor density and activation suggests a plausible biologic mechanism by which acute and chronic exercise training can decrease systemic levels of the IGF ligands through increased myocyte uptake. In response to exercise, there appears to be a physiologic capacity for both increases and decreases in systemic levels of IGF ligands; however, the determinants of this direction remain to be confirmed which reflects the current inconsistency of results. The local and systemic expression of the IGF axis appears inextricably related within the context of exercise. Quantification of intramuscular expression of the ligands, binding proteins, or receptors of the IGF axis was not included in any of the trials within the current review, limiting conclusions pertaining to the effects of exercise prescription. Future trials should endeavor to quantify the intramuscular expression of the IGF axis constituents in response to exercise concurrently with systemic changes to determine the exercise-induced response and any differential effects between systems [see the analysis of Lemmey and colleagues (61) for reference].