Fatigue status in this sample was confirmed with the Multidimensional Fatigue Symptom Inventory, with the fatigued group scoring significantly higher on the general fatigue scale (mean = 12.5, SD = 5.1) than the non-fatigued group (mean = 3.9, SD = 3.6; t(45) = −5.7, p < .0001). Demographic, medical, and treatment-related characteristics of fatigued and non-fatigued survivors are shown in . There were significant group differences in depressive symptoms (t(45) = −2.8, p = .007) and age (t(45) = 2.6, p = .01). None of the other group differences were significant (all ps > .10).
Demographic and treatment-related characteristics of study participants
Analysis of the IL1B −511 (C/T) polymorphism revealed a substantial over-representation of CC alleles among fatigued survivors and a substantial under-representation of TT alleles (). Increased prevalence of C vs. T nucleotides at IL1B −511 yielded a linear dose–response relationship (p = .008), and the prevalence of at least one cytosine at this position was substantially greater among fatigued participants (100%) than among non-fatigued (78.6%; 95% CI = 0.91–16.6, p = .007).
Fig. 1 Prevalence of polymorphisms in the interleukin-1 beta gene (IL1B) and the interleukin-6 gene (IL6) in fatigued and non-fatigued breast cancer survivors. Significant predictors of fatigue status included presence of at least one cytosine at IL1B −511 (more ...)
Fatigued breast cancer survivors showed an elevated occurrence of homozygosity for both the variant C allele and the wildtype G allele of the IL6 −174 (G/C) polymorphism, with approximately twofold greater representation of both GG and CC alleles in the fatigued group compared to non-fatigued controls (). Only 33.3% of fatigued survivors were heterozygous for IL6 −174, compared to 69.2% of non-fatigued controls (95% CI = 1.12–17.9; p = .027).
To determine whether IL1β −511 and IL6 −174 rSNPs each constitute independent predictors for post-cancer fatigue, we carried out multivariate logistic regression analyses predicting fatigue status from both polymorphisms in parallel. Results showed that each polymorphism was independently associated with fatigue status, with fatigue continuing to show significant association with homozygosity at IL6 −174 (p = .024) and the prevalence of cytosine at IL1B −511 (p = .021). Consistent results were obtained in analyses restricting fatigue to individuals who scored at or below 50 on the SF-36 vitality scale.
To determine whether associations between genotype and fatigue status were independent of other demographic and clinical confounds (i.e., age, ethnicity, menopausal status, body mass index, depressive symptoms, and cancer treatments), we first identified confounds that were significantly or nearly significantly associated with one or more of the measured polymorphisms. Analyses controlling for these variables (i.e., age, depressive symptoms, mastectomy, breast reconstruction, and breast irradiation) continued to show significant relationships between fatigue and the IL1B −511 polymorphism, with the exception that controlling for depressive symptoms resulted in a near-significant association between fatigue and cytosine frequency at IL1B −511 (p = .052). However, relationships between IL6 −174 homozygosity and fatigue were rendered non-significant by control for age, and by control for several treatment-related characteristics (all p > .05). Analyses controlling for ethnicity (white vs. non-white) continued to show significant associations between each genotype and fatigue status.
As previously reported, fatigued women in the parent study from which these women were drawn had elevated plasma levels of IL-1ra and soluble IL-6R relative to non-fatigued survivors, as well as elevations in monocyte intracellular production of IL-6 following ex vivo
LPS stimulation (Collado-Hidalgo et al., 2006
). Consistent with results from the previous study, fatigued women in this sample showed significantly higher levels of sIL-6R (p
= .028) and marginally significantly higher levels of IL-1ra (p
= .074). There were no significant differences in plasma levels of IL-6, again consistent with results from the parent study.
Thus, exploratory analyses were conducted to evaluate whether IL1B −511 and IL6 −174 rSNPs were associated with alterations in these parameters using Spearman rank correlation coefficients. We found significant correlations between circulating concentrations of IL-1ra and the frequency of G alleles at IL6 −174 (r = .35, p = .019), and between circulating concentrations of sIL-6R and the frequency of C alleles at IL1B −511 (r = .33, p = .027). There was also a trend towards an association between frequency of G alleles at IL6 −174 and increased ex vivo monocyte production of IL-6 (r = .26, p = .078). However, we found no association between fatigue genotypes and circulating levels of IL-6.