Is Leptin a Mediator of Adverse Prognostic Effects of Obesity in Breast Cancer?
http://www.100md.com
《临床肿瘤学》
the Department of Medicine, Department of Surgery, Division of Clinical Epidemiology at the Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto-Sunnybrook Regional Cancer Centre, St Michael's Hospital, University of Toronto, Toronto, Ontario, Canada
ABSTRACT
PURPOSE: Leptin, an adipocyte-derived cytokine that is elevated in obesity, has been associated with carcinogenesis, tumor migration and invasion, enhancement of angiogenesis, and increased aromatase activity. It has been suggested that leptin may mediate adverse prognostic effects of obesity in breast cancer.
PATIENTS AND METHODS: Four hundred seventy-one women with surgically resected T1-3, N0-1, M0 breast cancer were studied. Leptin was assayed in stored fasting blood specimens obtained before adjuvant therapy. Women were followed prospectively for distant disease-free survival (DDFS) and overall survival (OS).
RESULTS: Patients ranged from 26 to 74 years of age, and staging was as follows: T1 = 262, T2 = 151, T3 = 23, TX = 35, N0 = 323, and N1 = 148. Estrogen receptor was positive in 286 patients, and progesterone receptor was positive in 259 patients. One hundred forty-five patients received adjuvant chemotherapy, 146 received adjuvant tamoxifen, 46 received both, and 134 received neither. Mean leptin was 15.2 ± 10.1 ng/mL. Univariately, leptin was associated with OS (overall P = .049; P = .014 postmenopausal). Leptin was not associated with DDFS overall or in any menopausal subgroup (P .19). In multivariate Cox modeling, leptin was not significantly associated with DDFS or OS (P = .11 and 0.075, respectively). Adjustment for insulin or body mass index further reduced the association of leptin with outcome.
CONCLUSION: Although leptin is strongly correlated with obesity and insulin, we could not show that it is independently associated with prognosis in early-stage breast cancer. Because we cannot rule out modest prognostic effects, we recommend additional research to explore this potential association, particularly in postmenopausal women.
INTRODUCTION
Obesity is widely recognized as an adverse prognostic factor in breast cancer.1 In a recently reported prospective cohort study, we confirmed an adverse prognostic effect of obesity and identified insulin as a potential physiologic mediator of this prognostic effect.2 A role of insulin in breast cancer prognosis is biologically plausible, given reports that insulin receptors are overexpressed in the majority of breast cancers,3 that insulin receptor-alpha (a fetal insulin receptor associated with enhanced mitogenic signaling) is often present on breast cancers,4 and that insulin and insulin-like growth factor I (IGF-I) cross-react with their cognate receptors. Insulin also modifies levels of IGF-I, IGF-binding protein-3, and bioavailable estrogen, resulting in potential indirect effects on breast cancer outcomes. Whether insulin influences breast cancer prognosis through direct or indirect mechanisms remains unresolved.5,6
Rose et al recently suggested that "[t]he relationship between hyperinsulinism and a poor breast cancer prognosis reported by Goodwin et al is, at least in part, mediated by way of enhanced leptin production by adipose tissue and elevated levels of the hormone in obese patients."7 Leptin is a neuroendocrine hormone that is the product of the obesity gene, expressed primarily by adipocytes.8 Plasma leptin concentrations are a biomarker of obesity.9 Rose et al7 reviewed evidence that leptin is associated with stimulation of normal and tumor cell growth, tumor migration and invasion, enhancement of angiogenesis, and enhanced aromatase activity. These biologic actions could lead to worsened breast cancer prognosis.
In this report, we empirically examine the association of leptin with obesity and the prognostic effects of leptin in an existing cohort of breast cancer patients.
PATIENTS AND METHODS
Population Assembly
Details of recruitment are reported elsewhere.2 Briefly, 535 eligible and consenting women were enrolled onto the study between 1989 and 1996. Women were eligible if they were < 75 years of age (between 1989 and 1992, only premenopausal women were recruited) and had undergone a complete resection (lumpectomy with margins clear of invasive cancer or mastectomy) with axillary node dissection for previously untreated T1-3, N0-1, M0 breast cancer. Women were excluded if they had a prior malignancy, had a serious coexisting medical conditions (including type I or II diabetes or use of medications that could influence diet and lipids), were unable to speak English, or refused to provide informed consent. Fasting blood was collected from 520 of these women. Stored blood was available for leptin analysis on 471 women (no specimens remained after previous analyses for the remaining women). All participants provided written informed consent in accordance with approval granted by the Human Subjects Committee of the University of Toronto and participating institutions.
Measurement
Women underwent baseline measurements between four and 12 weeks postoperatively, before adjuvant therapy. Weight was measured by using a balance beam scale after a 12-hour overnight fast with the patient clothed in a hospital gown. Body mass index (BMI) was calculated as weight (kg)/height (m2). Fasting blood was collected into tubes containing EDTA anticoagulant and centrifuged immediately, and the plasma was stored at –70°C without thawing before leptin assays.
Pathologic characteristics of tumors (including hormone receptors) were abstracted from pathology reports. Hormone receptors were measured at participating institutions by using protein-binding or immunohistochemical assays according to the standard practice in place at each institution.
Insulin was measured on the automated Beckman-Coulter Access Immunoassay System (Beckman-Coulter Canada, Inc., Mississauga, Canada) using the manufacturer's two-epitope immunometric chemiluminescent method. Leptin was measured by using the Linco human leptin kit (catalog No. EZ HL-81k) solid-phase, enzyme-linked immunosorbent assay. The limit of sensitivity was 0.5 ng/mL. There is no cross reactivity with human insulin, proinsulin, IGF-I, IGF-II, or glucagon. The mean within-assay and between-assay coefficients of variation were 3.05% and 4.72%, respectively. Assays were performed in 2001, an average of 8 years after blood collection. There was no evidence that leptin degraded over time; mean levels (adjusted for age because only premenopausal women were recruited during the first 3 years of the study) were constant by calendar year of collection, reflecting storage over a range of 5 to 12 years (P [trend] was not significant).
Follow-Up
Participants were followed prospectively, and information on disease recurrence and death was abstracted from medical records. All medical events were reviewed in a standard fashion by P.J.G., who was blinded with respect to BMI and insulin and leptin levels. Seven women were lost to follow-up and censored at last contact. Recurrences were categorized by using criteria established by the National Cancer Institute of Canada Clinical Trials Group.
Statistical Analysis
Descriptive means, standard deviations, and/or distributions were generated for all study variables. Distributions of continuous variables were checked, and the following transformations were performed to reduce skewness and long tails: leptin was raised to the power of 0.25 and insulin to the power of –0.25; BMI was inverse. All statistical tests and model building were performed by using the transformed variables; the transformations were reversed for presentation of the results.
Spearman rank-correlation coefficients were calculated to examine correlations with leptin. Mean leptin levels across categories of tumor stage, nodal stage, tumor grade, and hormone receptors were examined and compared by using one-way analysis of variance F tests to examine the relationship of leptin to traditional prognostic factors.
Cox proportional-hazards models were used to examine the univariate effects of leptin on distant disease-free survival (DDFS) and overall survival (OS). Kaplan-Meier estimates of OS, with leptin categorized by splitting it at its quartiles, were calculated. Distant recurrences included those beyond the ipsilateral breast and axilla and did not include new primary cancers in the contralateral breast.
Multivariate prognostic analyses using Cox models were performed for leptin by adjusting for age at diagnosis (continuous), tumor stage (T2, T3, or TX v T1), nodal stage (N1 v N0), tumor grade (3 v 1 or 2), hormone receptor status (both negative v either estrogen receptor [ER] or progesterone receptor [PgR] positive or equivocal), adjuvant chemotherapy (yes v no), and adjuvant tamoxifen (yes v no). Nuclear grade and lymphatic invasion were not included, because they were missing in a nonrandom fashion in a large number of patients. In separate models, insulin and BMI (the latter modeled quadratically) were added to the set of adjusters. The possibility of a curvilinear relationship for leptin was explored by repeating the Cox models with leptin formulated as a quadratic function.
For the purposes of presentation of the prognostic effect sizes of leptin, hazard ratios (HRs), and 95% CIs were calculated from the Cox models using the 12.5th, 37.5th, 62.5th, and 87.5th percentiles of leptin as representative points. The P values reported were obtained from likelihood ratio 2 tests for leptin in the Cox models. All P values are two-tailed.
RESULTS
Study Population
As can be seen in Table 1, mean age was 50.7 ± 9.8 years. The majority of patients were premenopausal (54.1%). The majority had T1 (55.6%), N0 (68.6%), ER-positive (60.7%), PgR-positive (55.0%) tumors. Adjuvant chemotherapy was administered to 40.6% of the patients, and adjuvant hormone therapy (tamoxifen) was administered to 40.8% of the patients. Median follow-up (censoring patients at death) was 71.6 months (range, 1 to 135 months). Seventy-nine women experienced distant recurrences, and 49 died. All but three deaths were caused by breast cancer.
Leptin, BMI, and Related Factors
The mean leptin level ± standard deviation was 15.2 ± 10.1 ng/mL. As can be seen in Table 2, leptin was strongly correlated with BMI (Spearman's r = 0.81; P < .001). Leptin was also strongly correlated with fasting insulin (r = 0.64; P < .001). Leptin was moderately inversely correlated with IGF-binding protein-1 (r = –0.45; P < .001). Correlations with other related compounds including glucose and IGF-I were weaker.
Leptin and Tumor-Related Prognostic Variables
We examined the association of leptin levels with key tumor-related prognostic variables (Table 3). Leptin levels were significantly higher with higher T stage (P = .0015), higher tumor grade (P = .0023), or ER or PgR negativity (P = .047 and .080, respectively), all of which are associated with worse prognosis. However, there was no evidence that leptin levels were higher when cancer had spread to axillary lymph nodes (15.0 ± 9.7 in N0, 15.8 ± 10.9 in N1; P = .59).
Prognostic Associations of Leptin
Prognostic associations of leptin are listed in Table 4. In the first section, which provides results of univariate Cox models, it can be seen that high leptin levels were not significantly associated with an increased risk of distant recurrence (quartile 4 v 1 HR, 1.41 [95% CI, 0.85 to 2.35; P = .19]). This association was of borderline significance for OS (HR, 1.92; 95% CI, 1.00 to 3.69; P = .049). In the next section, it can be seen that adjustment for age, T stage, N stage, ER/PgR, tumor grade, and adjuvant therapy resulted in absence of a statistically significant effect on either DDFS or OS (observed HRs [95% CIs] for midpoints of upper v lower quartiles, 1.58 [0.90 to 2.79] and 1.95 [0.93 to 4.08], respectively). These observations suggest that leptin did not make significant independent contributions to DDFS or OS after consideration of tumor- and treatment-related variables. Finally, in the last section, prognostic effects of leptin after adjustment for traditional prognostic factors, adjuvant therapy, and insulin are shown. It can be seen that prognostic effects of leptin were not significant (quartile 4 v 1 HR, 1.26 [95% CI, 0.61 to 2.64; P = .53] for DDFS and 1.29 [95% CI, 0.49 to 3.37; P = .60] for OS). Inclusion of BMI (rather than insulin) as a covariate, along with tumor- and treatment-related variables, also resulted in nonsignificant prognostic effects of leptin on DDFS (P = .78) and OS (P = .52). OS curves for the four leptin quartiles are shown in Figure 1. Curves for the quartiles cross, and there is no evidence of a monotonic worsening of prognosis with increasing leptin quartile.
Because a curvilinear association of BMI with breast cancer outcomes was seen in this cohort2 and leptin was strongly correlated with BMI (r = 0.81), we investigated the possibility that leptin was associated with breast cancer outcomes in a curvilinear fashion. Leptin was modeled quadratically (squared) in univariate and multivariate Cox models. In univariate models, significant associations of leptin with DDFS and OS were seen (P = .004 and .0009, respectively). However, similar to the pattern seen when leptin was modeled in a linear fashion, when tumor- and treatment-related variables were included in these Cox models, the association of leptin with DDFS and OS was less significant (P = .12 and .054, respectively). Associations were nonsignificant when BMI and/or insulin were included in the models (all P > .20). Absence of a significant quadratic association of leptin with outcome is in keeping with known biologic effects of leptin,7 which do not provide a rationale for adverse prognostic effects of low levels of leptin.
Because the effect of obesity on breast cancer risk seems to vary with respect to menopausal status (obesity is associated with increased risk in postmenopausal women only),10 we explored whether prognostic effects of leptin differed according to menopausal status. There was no evidence of a significant interaction between leptin and menopausal status (premenopausal/perimenopausal v postmenopausal) in unadjusted (interaction P = .19) or adjusted (interaction P = .20, adjusted for the tumor- and treatment-related factors listed above) analyses. However, in subset analyses, leptin was significantly associated with OS in postmenopausal women (quartile 4 v 1 HR, 6.1; P = .014) but not in premenopausal or perimenopausal women (quartile 4 v 1 HR, 1.74; P = .21). As in our overall analyses, adjustment for T and N stage, tumor grade, ER/PgR, and adjuvant therapy resulted in absence of a significant effect in both premenopausal/perimenopausal and postmenopausal women (quartile 4 v 1 HRs, 1.63 [P = .32] and 3.35 [P = .15], respectively). The latter HR decreased to 1.63 (95% CI, 0.14 to 19.4) when BMI was included in the model. We also explored whether survival effects of leptin differed in women who received (or did not receive) systemic adjuvant therapy. There was no evidence of a significant interaction (univariate interaction, P = .92). Furthermore, subset analyses in those who did not receive systemic adjuvant therapy identified no significant effects even before adjustment for tumor- and treatment-related variables (all P > .10).
DISCUSSION
We have been unable to demonstrate a significant effect of leptin on outcome in early-stage breast cancer despite the fact that plasma leptin levels were strongly correlated with both BMI and insulin, factors that we previously showed to be related to breast cancer outcome. Plasma leptin levels were higher when some adverse prognostic factors (eg, advanced tumor stage, high tumor grade, and hormone receptor negativity) were present; however, they were not associated with spread of tumor to axillary lymph nodes. Even before adjustment for tumor-related factors, the association of leptin with DDFS was not significant, and the association with OS was of borderline significance. After adjustment for these factors, leptin was not significantly associated with distant recurrence, and the association with death was of borderline significance (P = .075). Of note, after adjustment for insulin, there was virtually no association of leptin with either distant recurrence or death. Similar associations were identified regardless of whether a linear or quadratic association of leptin with breast cancer outcomes was modeled in multivariate Cox models. Analyses exploring survival effects of leptin in menopausal subgroups suggested that univariate (but not multivariate) effects of leptin might be greatest in postmenopausal women. This subset should be investigated more fully in future research. Additionally, because systemic adjuvant therapy (chemotherapy, hormone therapy) is more commonly used today than during the recruitment phase of this study, additional investigation of prognostic effects of leptin in women receiving modern adjuvant therapy, including anthracycline-based chemotherapy and aromatase inhibitors, is warranted.
Although we studied almost 500 women and have been able to demonstrate adverse prognostic effects of insulin and BMI in this population,2 the size of our study may have limited our ability to identify prognostic effects of leptin. We had 80% power to detect an HR of 2.3 for DDFS and 2.9 for OS in our study population. Thus, we cannot rule out the presence of more modest survival effects. Replication of our work in a larger cohort is recommended to rigorously investigate modest prognostic effects of leptin (eg, HRs in the range of 1.5 to 2.5) and to explore potential effects in postmenopausal women.
High leptin levels were associated with advanced tumor stage, higher tumor grade, and hormone receptor negativity but not axillary nodal metastases. It is possible that high leptin levels before breast cancer diagnosis may lead to a more aggressive breast cancer phenotype; however, high leptin levels do not seem to be associated with metastases to axillary lymph nodes or to distant sites. These observations require replication; however, they suggest that leptin may contribute to the development of an aggressive malignant phenotype rather than to the development of the potential for metastatic spread.
To our knowledge, there have been no previous reports investigating prognostic effects of leptin in breast cancer. However, there have been three studies examining the association of leptin with breast cancer risk. Because the association of obesity with breast cancer risk (adverse effect present only in postmenopausal but not premenopausal women)11 differs from the association of obesity with breast cancer prognosis (adverse effect present in both premenopausal and postmenopausal women),1 caution should be exercised in extrapolating studies of the role of leptin in breast cancer risk to its role in breast cancer prognosis. Four studies12-15 have examined the association of leptin with breast cancer risk. One study12 involving 75 women with newly diagnosed breast cancer and 75 healthy controls reported an inverse association of leptin with risk in premenopausal but not postmenopausal women. Another study13 involving 58 women with breast cancer of all stages and 58 healthy controls reported higher levels of leptin in breast cancer subjects, an effect that seemed to result from the use of tamoxifen. In the final two studies, leptin was not associated with risk of premenopausal in situ breast cancer in 83 cases and 69 healthy controls,14 nor was it associated with risk of postmenopausal invasive breast cancer in 149 cases and 258 healthy controls.15 Overall, these results provide little support for a clinically important role of circulating leptin in the development of breast cancer.
Although an adverse prognostic effect of leptin is biologically plausible, as reviewed by Rose et al,7 we have little empiric evidence to indicate that leptin exerts a major adverse prognostic effect in early-stage breast cancer or that it mediates adverse prognostic effects of obesity. As noted above, we were unable to exclude modest prognostic effects, because our sample size was restricted to 471 women. Furthermore, our exploratory observations in postmenopausal women warrant additional study. In contrast, in this same cohort we previously identified significant adversely prognostic effects of both BMI and insulin.2 The association of insulin with outcome was strongest (risk of death tripled) in women whose insulin levels were in the highest versus the lowest quartile; thus, in this cohort insulin seems to be the factor primarily associated with adverse breast cancer outcomes. Obesity (high BMI) was correlated with insulin (r = 0.59) and may contribute to hyperinsulinemia as part of the insulin-resistance syndrome.16 Although leptin is associated with both insulin and BMI, we could not demonstrate prognostic effects that were similar to those of insulin even before adjustment for insulin or BMI. Our results should be replicated in other data sets and include larger groups of postmenopausal women. Unless new evidence becomes available indicating a significant role for leptin in breast cancer outcomes, we would recommend that intervention research to improve breast cancer outcome focus on reduction in insulin levels either through pharmacologic means or reduction in obesity.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflict of interest.
NOTES
Funded by the Canadian Breast Cancer Research Alliance and the Medical Research Council of Canada (currently the Canadian Institutes of Health Research).
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Chlebowski RT, Aiello E, McTiernan A: Weight loss in breast cancer patient management. J Clin Oncol 20:1128-1143, 2002
Goodwin PJ, Ennis M, Pritchard KI, et al: Fasting insulin and outcome in early-stage breast cancer: Results of a prospective cohort study. J Clin Oncol 20:42-51, 2002
Papa V, Belfiore A: Insulin receptors in breast cancer: Biological and clinical role. J Endocrinol Invest 19:324-333, 1996
Frasca F, Pandini G, Scalia P, et al: Insulin receptor isoform A, a newly recognized, high-affinity insulin-like growth factor II receptor in fetal and cancer cells. Mol Cell Biol 19:3278-3288, 1999
Boyd DB: Insulin and cancer. Integr Cancer Ther 2:315-329, 2003
Sandhu MS, Dunger DB, Giovannucci EL: Insulin, insulin-like growth factor-I (IGF-I), IGF binding proteins, their biologic interactions, and colorectal cancer. J Natl Cancer Inst 94:972-980, 2002
Rose DP, Gilhooly EM, Nixon DW: Adverse effects of obesity on breast cancer prognosis, and the biological actions of leptin (review). Int J Oncol 21:1285-1292, 2002
Huang L, Li C: Leptin: A multifunctional hormone. Cell Res 10:81-92, 2000
Fung TT, Rimm EB, Spiegelman D, et al: Association between dietary patterns and plasma biomarkers of obesity and cardiovascular disease risk. Am J Clin Nutr 73:61-67, 2001
Harvie M, Hooper L, Howell AH: Central obesity and breast cancer risk: A systematic review. Obes Rev 4:157-173, 2003
Stephenson GD, Rose DP: Breast cancer and obesity: An update. Nutr Cancer 45:1-16, 2003
Petridou E, Papadiamantis Y, Markopoulos C, et al: Leptin and insulin growth factor I in relation to breast cancer (Greece). Cancer Causes Control 11:383-388, 2000
Ozet A, Arpaci F, Ilker Yilmaz M, et al: Effects of tamoxifen on serum leptin level in patients with breast cancer. Jpn J Clin Oncol 31:424-427, 2001
Mantzoros CS, Bolhke K, Moschos S, et al: Leptin in relation to carcinoma in situ of the breast: A study of pre-menopausal cases and controls. Int J Cancer 80:523-526, 1999
Stattin P, Soderberg S, Biessy C, et al: Plasma leptin and breast cancer risk: A prospective study in northern Sweden. Breast Cancer Res Treat 86:191-196, 2004
Doelle GC: The clinical picture of metabolic syndrome. An update on this complex of conditions and risk factors. Postgrad Med 116:30-32, 35-38, 2004(Pamela J. Goodwin, Margue)
ABSTRACT
PURPOSE: Leptin, an adipocyte-derived cytokine that is elevated in obesity, has been associated with carcinogenesis, tumor migration and invasion, enhancement of angiogenesis, and increased aromatase activity. It has been suggested that leptin may mediate adverse prognostic effects of obesity in breast cancer.
PATIENTS AND METHODS: Four hundred seventy-one women with surgically resected T1-3, N0-1, M0 breast cancer were studied. Leptin was assayed in stored fasting blood specimens obtained before adjuvant therapy. Women were followed prospectively for distant disease-free survival (DDFS) and overall survival (OS).
RESULTS: Patients ranged from 26 to 74 years of age, and staging was as follows: T1 = 262, T2 = 151, T3 = 23, TX = 35, N0 = 323, and N1 = 148. Estrogen receptor was positive in 286 patients, and progesterone receptor was positive in 259 patients. One hundred forty-five patients received adjuvant chemotherapy, 146 received adjuvant tamoxifen, 46 received both, and 134 received neither. Mean leptin was 15.2 ± 10.1 ng/mL. Univariately, leptin was associated with OS (overall P = .049; P = .014 postmenopausal). Leptin was not associated with DDFS overall or in any menopausal subgroup (P .19). In multivariate Cox modeling, leptin was not significantly associated with DDFS or OS (P = .11 and 0.075, respectively). Adjustment for insulin or body mass index further reduced the association of leptin with outcome.
CONCLUSION: Although leptin is strongly correlated with obesity and insulin, we could not show that it is independently associated with prognosis in early-stage breast cancer. Because we cannot rule out modest prognostic effects, we recommend additional research to explore this potential association, particularly in postmenopausal women.
INTRODUCTION
Obesity is widely recognized as an adverse prognostic factor in breast cancer.1 In a recently reported prospective cohort study, we confirmed an adverse prognostic effect of obesity and identified insulin as a potential physiologic mediator of this prognostic effect.2 A role of insulin in breast cancer prognosis is biologically plausible, given reports that insulin receptors are overexpressed in the majority of breast cancers,3 that insulin receptor-alpha (a fetal insulin receptor associated with enhanced mitogenic signaling) is often present on breast cancers,4 and that insulin and insulin-like growth factor I (IGF-I) cross-react with their cognate receptors. Insulin also modifies levels of IGF-I, IGF-binding protein-3, and bioavailable estrogen, resulting in potential indirect effects on breast cancer outcomes. Whether insulin influences breast cancer prognosis through direct or indirect mechanisms remains unresolved.5,6
Rose et al recently suggested that "[t]he relationship between hyperinsulinism and a poor breast cancer prognosis reported by Goodwin et al is, at least in part, mediated by way of enhanced leptin production by adipose tissue and elevated levels of the hormone in obese patients."7 Leptin is a neuroendocrine hormone that is the product of the obesity gene, expressed primarily by adipocytes.8 Plasma leptin concentrations are a biomarker of obesity.9 Rose et al7 reviewed evidence that leptin is associated with stimulation of normal and tumor cell growth, tumor migration and invasion, enhancement of angiogenesis, and enhanced aromatase activity. These biologic actions could lead to worsened breast cancer prognosis.
In this report, we empirically examine the association of leptin with obesity and the prognostic effects of leptin in an existing cohort of breast cancer patients.
PATIENTS AND METHODS
Population Assembly
Details of recruitment are reported elsewhere.2 Briefly, 535 eligible and consenting women were enrolled onto the study between 1989 and 1996. Women were eligible if they were < 75 years of age (between 1989 and 1992, only premenopausal women were recruited) and had undergone a complete resection (lumpectomy with margins clear of invasive cancer or mastectomy) with axillary node dissection for previously untreated T1-3, N0-1, M0 breast cancer. Women were excluded if they had a prior malignancy, had a serious coexisting medical conditions (including type I or II diabetes or use of medications that could influence diet and lipids), were unable to speak English, or refused to provide informed consent. Fasting blood was collected from 520 of these women. Stored blood was available for leptin analysis on 471 women (no specimens remained after previous analyses for the remaining women). All participants provided written informed consent in accordance with approval granted by the Human Subjects Committee of the University of Toronto and participating institutions.
Measurement
Women underwent baseline measurements between four and 12 weeks postoperatively, before adjuvant therapy. Weight was measured by using a balance beam scale after a 12-hour overnight fast with the patient clothed in a hospital gown. Body mass index (BMI) was calculated as weight (kg)/height (m2). Fasting blood was collected into tubes containing EDTA anticoagulant and centrifuged immediately, and the plasma was stored at –70°C without thawing before leptin assays.
Pathologic characteristics of tumors (including hormone receptors) were abstracted from pathology reports. Hormone receptors were measured at participating institutions by using protein-binding or immunohistochemical assays according to the standard practice in place at each institution.
Insulin was measured on the automated Beckman-Coulter Access Immunoassay System (Beckman-Coulter Canada, Inc., Mississauga, Canada) using the manufacturer's two-epitope immunometric chemiluminescent method. Leptin was measured by using the Linco human leptin kit (catalog No. EZ HL-81k) solid-phase, enzyme-linked immunosorbent assay. The limit of sensitivity was 0.5 ng/mL. There is no cross reactivity with human insulin, proinsulin, IGF-I, IGF-II, or glucagon. The mean within-assay and between-assay coefficients of variation were 3.05% and 4.72%, respectively. Assays were performed in 2001, an average of 8 years after blood collection. There was no evidence that leptin degraded over time; mean levels (adjusted for age because only premenopausal women were recruited during the first 3 years of the study) were constant by calendar year of collection, reflecting storage over a range of 5 to 12 years (P [trend] was not significant).
Follow-Up
Participants were followed prospectively, and information on disease recurrence and death was abstracted from medical records. All medical events were reviewed in a standard fashion by P.J.G., who was blinded with respect to BMI and insulin and leptin levels. Seven women were lost to follow-up and censored at last contact. Recurrences were categorized by using criteria established by the National Cancer Institute of Canada Clinical Trials Group.
Statistical Analysis
Descriptive means, standard deviations, and/or distributions were generated for all study variables. Distributions of continuous variables were checked, and the following transformations were performed to reduce skewness and long tails: leptin was raised to the power of 0.25 and insulin to the power of –0.25; BMI was inverse. All statistical tests and model building were performed by using the transformed variables; the transformations were reversed for presentation of the results.
Spearman rank-correlation coefficients were calculated to examine correlations with leptin. Mean leptin levels across categories of tumor stage, nodal stage, tumor grade, and hormone receptors were examined and compared by using one-way analysis of variance F tests to examine the relationship of leptin to traditional prognostic factors.
Cox proportional-hazards models were used to examine the univariate effects of leptin on distant disease-free survival (DDFS) and overall survival (OS). Kaplan-Meier estimates of OS, with leptin categorized by splitting it at its quartiles, were calculated. Distant recurrences included those beyond the ipsilateral breast and axilla and did not include new primary cancers in the contralateral breast.
Multivariate prognostic analyses using Cox models were performed for leptin by adjusting for age at diagnosis (continuous), tumor stage (T2, T3, or TX v T1), nodal stage (N1 v N0), tumor grade (3 v 1 or 2), hormone receptor status (both negative v either estrogen receptor [ER] or progesterone receptor [PgR] positive or equivocal), adjuvant chemotherapy (yes v no), and adjuvant tamoxifen (yes v no). Nuclear grade and lymphatic invasion were not included, because they were missing in a nonrandom fashion in a large number of patients. In separate models, insulin and BMI (the latter modeled quadratically) were added to the set of adjusters. The possibility of a curvilinear relationship for leptin was explored by repeating the Cox models with leptin formulated as a quadratic function.
For the purposes of presentation of the prognostic effect sizes of leptin, hazard ratios (HRs), and 95% CIs were calculated from the Cox models using the 12.5th, 37.5th, 62.5th, and 87.5th percentiles of leptin as representative points. The P values reported were obtained from likelihood ratio 2 tests for leptin in the Cox models. All P values are two-tailed.
RESULTS
Study Population
As can be seen in Table 1, mean age was 50.7 ± 9.8 years. The majority of patients were premenopausal (54.1%). The majority had T1 (55.6%), N0 (68.6%), ER-positive (60.7%), PgR-positive (55.0%) tumors. Adjuvant chemotherapy was administered to 40.6% of the patients, and adjuvant hormone therapy (tamoxifen) was administered to 40.8% of the patients. Median follow-up (censoring patients at death) was 71.6 months (range, 1 to 135 months). Seventy-nine women experienced distant recurrences, and 49 died. All but three deaths were caused by breast cancer.
Leptin, BMI, and Related Factors
The mean leptin level ± standard deviation was 15.2 ± 10.1 ng/mL. As can be seen in Table 2, leptin was strongly correlated with BMI (Spearman's r = 0.81; P < .001). Leptin was also strongly correlated with fasting insulin (r = 0.64; P < .001). Leptin was moderately inversely correlated with IGF-binding protein-1 (r = –0.45; P < .001). Correlations with other related compounds including glucose and IGF-I were weaker.
Leptin and Tumor-Related Prognostic Variables
We examined the association of leptin levels with key tumor-related prognostic variables (Table 3). Leptin levels were significantly higher with higher T stage (P = .0015), higher tumor grade (P = .0023), or ER or PgR negativity (P = .047 and .080, respectively), all of which are associated with worse prognosis. However, there was no evidence that leptin levels were higher when cancer had spread to axillary lymph nodes (15.0 ± 9.7 in N0, 15.8 ± 10.9 in N1; P = .59).
Prognostic Associations of Leptin
Prognostic associations of leptin are listed in Table 4. In the first section, which provides results of univariate Cox models, it can be seen that high leptin levels were not significantly associated with an increased risk of distant recurrence (quartile 4 v 1 HR, 1.41 [95% CI, 0.85 to 2.35; P = .19]). This association was of borderline significance for OS (HR, 1.92; 95% CI, 1.00 to 3.69; P = .049). In the next section, it can be seen that adjustment for age, T stage, N stage, ER/PgR, tumor grade, and adjuvant therapy resulted in absence of a statistically significant effect on either DDFS or OS (observed HRs [95% CIs] for midpoints of upper v lower quartiles, 1.58 [0.90 to 2.79] and 1.95 [0.93 to 4.08], respectively). These observations suggest that leptin did not make significant independent contributions to DDFS or OS after consideration of tumor- and treatment-related variables. Finally, in the last section, prognostic effects of leptin after adjustment for traditional prognostic factors, adjuvant therapy, and insulin are shown. It can be seen that prognostic effects of leptin were not significant (quartile 4 v 1 HR, 1.26 [95% CI, 0.61 to 2.64; P = .53] for DDFS and 1.29 [95% CI, 0.49 to 3.37; P = .60] for OS). Inclusion of BMI (rather than insulin) as a covariate, along with tumor- and treatment-related variables, also resulted in nonsignificant prognostic effects of leptin on DDFS (P = .78) and OS (P = .52). OS curves for the four leptin quartiles are shown in Figure 1. Curves for the quartiles cross, and there is no evidence of a monotonic worsening of prognosis with increasing leptin quartile.
Because a curvilinear association of BMI with breast cancer outcomes was seen in this cohort2 and leptin was strongly correlated with BMI (r = 0.81), we investigated the possibility that leptin was associated with breast cancer outcomes in a curvilinear fashion. Leptin was modeled quadratically (squared) in univariate and multivariate Cox models. In univariate models, significant associations of leptin with DDFS and OS were seen (P = .004 and .0009, respectively). However, similar to the pattern seen when leptin was modeled in a linear fashion, when tumor- and treatment-related variables were included in these Cox models, the association of leptin with DDFS and OS was less significant (P = .12 and .054, respectively). Associations were nonsignificant when BMI and/or insulin were included in the models (all P > .20). Absence of a significant quadratic association of leptin with outcome is in keeping with known biologic effects of leptin,7 which do not provide a rationale for adverse prognostic effects of low levels of leptin.
Because the effect of obesity on breast cancer risk seems to vary with respect to menopausal status (obesity is associated with increased risk in postmenopausal women only),10 we explored whether prognostic effects of leptin differed according to menopausal status. There was no evidence of a significant interaction between leptin and menopausal status (premenopausal/perimenopausal v postmenopausal) in unadjusted (interaction P = .19) or adjusted (interaction P = .20, adjusted for the tumor- and treatment-related factors listed above) analyses. However, in subset analyses, leptin was significantly associated with OS in postmenopausal women (quartile 4 v 1 HR, 6.1; P = .014) but not in premenopausal or perimenopausal women (quartile 4 v 1 HR, 1.74; P = .21). As in our overall analyses, adjustment for T and N stage, tumor grade, ER/PgR, and adjuvant therapy resulted in absence of a significant effect in both premenopausal/perimenopausal and postmenopausal women (quartile 4 v 1 HRs, 1.63 [P = .32] and 3.35 [P = .15], respectively). The latter HR decreased to 1.63 (95% CI, 0.14 to 19.4) when BMI was included in the model. We also explored whether survival effects of leptin differed in women who received (or did not receive) systemic adjuvant therapy. There was no evidence of a significant interaction (univariate interaction, P = .92). Furthermore, subset analyses in those who did not receive systemic adjuvant therapy identified no significant effects even before adjustment for tumor- and treatment-related variables (all P > .10).
DISCUSSION
We have been unable to demonstrate a significant effect of leptin on outcome in early-stage breast cancer despite the fact that plasma leptin levels were strongly correlated with both BMI and insulin, factors that we previously showed to be related to breast cancer outcome. Plasma leptin levels were higher when some adverse prognostic factors (eg, advanced tumor stage, high tumor grade, and hormone receptor negativity) were present; however, they were not associated with spread of tumor to axillary lymph nodes. Even before adjustment for tumor-related factors, the association of leptin with DDFS was not significant, and the association with OS was of borderline significance. After adjustment for these factors, leptin was not significantly associated with distant recurrence, and the association with death was of borderline significance (P = .075). Of note, after adjustment for insulin, there was virtually no association of leptin with either distant recurrence or death. Similar associations were identified regardless of whether a linear or quadratic association of leptin with breast cancer outcomes was modeled in multivariate Cox models. Analyses exploring survival effects of leptin in menopausal subgroups suggested that univariate (but not multivariate) effects of leptin might be greatest in postmenopausal women. This subset should be investigated more fully in future research. Additionally, because systemic adjuvant therapy (chemotherapy, hormone therapy) is more commonly used today than during the recruitment phase of this study, additional investigation of prognostic effects of leptin in women receiving modern adjuvant therapy, including anthracycline-based chemotherapy and aromatase inhibitors, is warranted.
Although we studied almost 500 women and have been able to demonstrate adverse prognostic effects of insulin and BMI in this population,2 the size of our study may have limited our ability to identify prognostic effects of leptin. We had 80% power to detect an HR of 2.3 for DDFS and 2.9 for OS in our study population. Thus, we cannot rule out the presence of more modest survival effects. Replication of our work in a larger cohort is recommended to rigorously investigate modest prognostic effects of leptin (eg, HRs in the range of 1.5 to 2.5) and to explore potential effects in postmenopausal women.
High leptin levels were associated with advanced tumor stage, higher tumor grade, and hormone receptor negativity but not axillary nodal metastases. It is possible that high leptin levels before breast cancer diagnosis may lead to a more aggressive breast cancer phenotype; however, high leptin levels do not seem to be associated with metastases to axillary lymph nodes or to distant sites. These observations require replication; however, they suggest that leptin may contribute to the development of an aggressive malignant phenotype rather than to the development of the potential for metastatic spread.
To our knowledge, there have been no previous reports investigating prognostic effects of leptin in breast cancer. However, there have been three studies examining the association of leptin with breast cancer risk. Because the association of obesity with breast cancer risk (adverse effect present only in postmenopausal but not premenopausal women)11 differs from the association of obesity with breast cancer prognosis (adverse effect present in both premenopausal and postmenopausal women),1 caution should be exercised in extrapolating studies of the role of leptin in breast cancer risk to its role in breast cancer prognosis. Four studies12-15 have examined the association of leptin with breast cancer risk. One study12 involving 75 women with newly diagnosed breast cancer and 75 healthy controls reported an inverse association of leptin with risk in premenopausal but not postmenopausal women. Another study13 involving 58 women with breast cancer of all stages and 58 healthy controls reported higher levels of leptin in breast cancer subjects, an effect that seemed to result from the use of tamoxifen. In the final two studies, leptin was not associated with risk of premenopausal in situ breast cancer in 83 cases and 69 healthy controls,14 nor was it associated with risk of postmenopausal invasive breast cancer in 149 cases and 258 healthy controls.15 Overall, these results provide little support for a clinically important role of circulating leptin in the development of breast cancer.
Although an adverse prognostic effect of leptin is biologically plausible, as reviewed by Rose et al,7 we have little empiric evidence to indicate that leptin exerts a major adverse prognostic effect in early-stage breast cancer or that it mediates adverse prognostic effects of obesity. As noted above, we were unable to exclude modest prognostic effects, because our sample size was restricted to 471 women. Furthermore, our exploratory observations in postmenopausal women warrant additional study. In contrast, in this same cohort we previously identified significant adversely prognostic effects of both BMI and insulin.2 The association of insulin with outcome was strongest (risk of death tripled) in women whose insulin levels were in the highest versus the lowest quartile; thus, in this cohort insulin seems to be the factor primarily associated with adverse breast cancer outcomes. Obesity (high BMI) was correlated with insulin (r = 0.59) and may contribute to hyperinsulinemia as part of the insulin-resistance syndrome.16 Although leptin is associated with both insulin and BMI, we could not demonstrate prognostic effects that were similar to those of insulin even before adjustment for insulin or BMI. Our results should be replicated in other data sets and include larger groups of postmenopausal women. Unless new evidence becomes available indicating a significant role for leptin in breast cancer outcomes, we would recommend that intervention research to improve breast cancer outcome focus on reduction in insulin levels either through pharmacologic means or reduction in obesity.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflict of interest.
NOTES
Funded by the Canadian Breast Cancer Research Alliance and the Medical Research Council of Canada (currently the Canadian Institutes of Health Research).
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Chlebowski RT, Aiello E, McTiernan A: Weight loss in breast cancer patient management. J Clin Oncol 20:1128-1143, 2002
Goodwin PJ, Ennis M, Pritchard KI, et al: Fasting insulin and outcome in early-stage breast cancer: Results of a prospective cohort study. J Clin Oncol 20:42-51, 2002
Papa V, Belfiore A: Insulin receptors in breast cancer: Biological and clinical role. J Endocrinol Invest 19:324-333, 1996
Frasca F, Pandini G, Scalia P, et al: Insulin receptor isoform A, a newly recognized, high-affinity insulin-like growth factor II receptor in fetal and cancer cells. Mol Cell Biol 19:3278-3288, 1999
Boyd DB: Insulin and cancer. Integr Cancer Ther 2:315-329, 2003
Sandhu MS, Dunger DB, Giovannucci EL: Insulin, insulin-like growth factor-I (IGF-I), IGF binding proteins, their biologic interactions, and colorectal cancer. J Natl Cancer Inst 94:972-980, 2002
Rose DP, Gilhooly EM, Nixon DW: Adverse effects of obesity on breast cancer prognosis, and the biological actions of leptin (review). Int J Oncol 21:1285-1292, 2002
Huang L, Li C: Leptin: A multifunctional hormone. Cell Res 10:81-92, 2000
Fung TT, Rimm EB, Spiegelman D, et al: Association between dietary patterns and plasma biomarkers of obesity and cardiovascular disease risk. Am J Clin Nutr 73:61-67, 2001
Harvie M, Hooper L, Howell AH: Central obesity and breast cancer risk: A systematic review. Obes Rev 4:157-173, 2003
Stephenson GD, Rose DP: Breast cancer and obesity: An update. Nutr Cancer 45:1-16, 2003
Petridou E, Papadiamantis Y, Markopoulos C, et al: Leptin and insulin growth factor I in relation to breast cancer (Greece). Cancer Causes Control 11:383-388, 2000
Ozet A, Arpaci F, Ilker Yilmaz M, et al: Effects of tamoxifen on serum leptin level in patients with breast cancer. Jpn J Clin Oncol 31:424-427, 2001
Mantzoros CS, Bolhke K, Moschos S, et al: Leptin in relation to carcinoma in situ of the breast: A study of pre-menopausal cases and controls. Int J Cancer 80:523-526, 1999
Stattin P, Soderberg S, Biessy C, et al: Plasma leptin and breast cancer risk: A prospective study in northern Sweden. Breast Cancer Res Treat 86:191-196, 2004
Doelle GC: The clinical picture of metabolic syndrome. An update on this complex of conditions and risk factors. Postgrad Med 116:30-32, 35-38, 2004(Pamela J. Goodwin, Margue)