In the current study 23% of the attending women had PF, i.e. they were diagnosed with CF at 2.5–7 years after their BC treatment (T1) and then again 2.5–3 years later (T2). CF was diagnosed in 33% of the attending women at T1 and in 39% at T2. However, these proportions do not always refer to the same individuals. While 10% had resolving fatigue with CF only at T1, 16% were incident cases with CF only at T2. Thus, our data demonstrate that CF after BC treatment is not necessarily a constant condition.
Current psychological distress, treatment-area related discomfort and BMI ≥ 30 were associated with CF at T1 and were also significant predictors for PF in the multiple analyses. In our sample an increased leukocyte count (still within the reference range) was also associated with PF. While respiratory symptoms and a higher hsCRP-level were related to CF in the cross-sectional study at T1, these factors did not significantly predict PF in the multiple model. Further, individuals with PF had significantly higher total fatigue scores at both assessments compared to the scores of subjects with CF only at either time point.
Our finding that about one third of the women had CF at each assessment point is in line with previously published results [1, 11]. Regarding the persistence of fatigue, one study reported that 19% of relapse-free women were severely fatigued both 29 months post-BC treatment and 2 years thereafter [30], while another reported that 21% of the BCSs surveyed 5–10 years after diagnosis had persistent fatigue [11]. The authors of the latter paper proposed that within the 1/3 of BCSs reporting fatigue at any time, a subgroup of women experienced “more debilitating fatigue symptoms” [11]. This is in accordance with our findings that subjects with PF had higher total fatigue scores at both assessment points compared to those with transient CF.
The findings of seeking help for mental problems prior to the BC and current psychological distress being associated with both CF and PF are in concordance with this study’s hypothesis and previous research [1, 6, 11, 15, 16, 19, 31]. Additionally, discomfort in the BC-treated area, representing both pain and/or fibrosis, was associated with CF and PF. Bower et al demonstrated that bodily pain was related to fatigue in women assessed up to 5 years post-BC treatment, and it was also a marginal predictor of fatigue at follow-up 5 years later [1, 11]. However, the location of pain was not specified. In another study, arm and shoulder pain was associated with fatigue post-BC treatment [18].
In addition to exploring self-reported factors, we also evaluated several clinical variables as possible predictors for CF and PF. We found that the median weight increase in women with CF was twice as large as in those without CF. In line with previous research [6, 7] we found that high post-BC BMI was associated with CF and a significant predictor of PF. Thus, efforts to reduce obesity in BC-treated women may contribute to lower their risk for developing persistent fatigue symptoms. The association between increased BMI and PF could also indicate a common underlying mechanism not explored in the present study design.
We observed a strong association between high hsCRP-level and treatment-area related discomfort. Therefore, only treatment-area related discomfort was entered into our final multiple logistic regression model and remained significantly associated with PF. In the univariate analysis hsCRP-level was a significant predictor of PF, with higher levels being associated with PF. Further, increased leukocyte count was associated with PF. Additionally, our data show that women with CF had higher levels of hsCRP compared to subjects without CF. Alexander et al also reported increased leukocyte count and CRP in relapse-free women with fatigue post-BC treatment, suggesting that a prolonged inflammatory response might underlie persistent fatigue [10]. Our data may also imply that an activated immune system plays a role in CF and PF, which is supported by gene expression analyses on a subsample of the women attending the present study [23]. However, a possible association between fatigue and inflammatory markers should be addressed in future studies.
We found no support for our hypothesis that young age at BC diagnosis was associated with CF or PF. Neither did we find that a diagnosis of hypothyroidism or an elevated TSH-level impacted on the women’s fatigue status. These results are in agreement with earlier research stating that abnormal thyroid function is not related to fatigue post-BC treatment [10].
We could not demonstrate any association between CF and PF and the BC treatment. However, all the included women were treated for stage II/III disease with postoperative radiotherapy, and most of them also received hormones and/or chemotherapy. Thus, our study was not optimally designed to assess the impact of different treatments upon fatigue.
A limitation of the current study is that information about the women’s fatigue status pre-BC diagnosis was not available. Therefore, we cannot conclude that fatigue was specifically related to BC. However, given the age-span relevant for the current study, CF in the general Norwegian population range from 12 to 22% [32]. Thus, CF is more prevalent in women after multimodal BC treatment than in the general population.
Even though not significant, our data indicate that the proportion of women with CF increased during the study’s time span. The women included in the current study had a higher systemic treatment burden than the non-responders. Therefore, a response bias caused by including subjects with more subjective symptoms cannot be excluded. Also in the general population the prevalence of fatigue increases with increasing age, other co-morbidities and strains of life [32]. Therefore, the observed increment in CF might be due to other factors than the BC or the BC treatment.
Also, having CF at both assessment points do not necessarily imply that the fatigue condition has been persistent during the whole observation period. However, previous papers on fatigue in BCSs used a similar approach by defining individuals with persistent fatigue as “survivors who reported significant fatigue on at least to occasions after cancer diagnosis and treatment” [21, 22, 33].
Definite information about the women’s pre-BC mental health status was not available and women visiting a psychologist/psychiatrist before their BC diagnosis were registered as having received “previous treatment for mental problems”. The survey run regularly by the Statistics Norway’s Health Interview Survey (http://www.ssb.no) reports that 4% of Norwegian females visited a psychologist/psychiatrist during the past 12 months in 1998. Thus, compared to the general Norwegian population, a considerably higher proportion (20%) of the women with CF had consulted a psychologist/psychiatrist prior to their BC diagnosis.
A recent review confirmed the association between fatigue and depression and anxiety, but the authors concluded that “directionality needs to be better delineated in longitudinal studies” [34]. Also in our study, directionality between CF and the associated factors is difficult to judge as data on the attending women’s pre-BC fatigue status were not available. However, we still consider our findings clinically important because most of the factors found at the first evaluation to be associated with the persistence of fatigue are possible targets for interventions.
Each woman was clinically examined by one oncologist. Independent oncologists examining the same individual could have allowed for estimation of the reliability of the ratings of fibrosis. However, the examination was performed by a limited number of experienced oncologists with a common scoring agreement supporting the reliability of this assessment.
Due to the relatively high number of study participants it was feasible to test multiple predictors possibly associated with PF. However, we did not have enough subjects to fit a model testing possible interactions between the variables.
Strengths of the current study are related to the long follow-up time and the longitudinal design enabling exploration of the course of CF and identification of individuals with PF as those with CF at both assessment points. Additional strength of the present study is the use of the FQ. Compared to other commonly used questionnaires assessing fatigue, the FQ has the advantage that the duration of fatigue is queried. Only subjects with symptom duration ≥6 months were defined as having CF, thus separating individuals with transient fatigue from those with more persistent symptoms. Finally, due to the unselective recruitment of patients to the hospital during the treatment period (1998–2002) and the standardized treatment they received based upon the existing national guidelines at that time, we consider the sample as representative for Norwegian women treated for BC stage II/III during the relevant treatment period.