You are here

A prognostic tool to predict fatigue in women with early-stage breast cancer undergoing radiotherapy

The Breast, 4, 22, pages 504 - 509

Abstract

Background

Fatigue during and after radiotherapy impacts negatively on normal functioning and quality of life. A pre-treatment estimate of the risk of fatigue would facilitate the targeting of timely interventions to limit consequential behavioural symptoms arising. We have developed a prognostic tool to predict the risk of fatigue in women with early-stage breast cancer undergoing radiotherapy.

Methods

Socio-demographic, clinical and self-reported characteristics were recorded for 100 women prescribed adjuvant radiotherapy for stages Tis–T2N1 breast cancer. Multiple logistic regression was used to develop a parsimonious prognostic model. The performance of the model when predicting fatigue for individuals not in the study was estimated by a leave-one-out cross-validation. A statistical weighting was assigned to the model variables to render a Fatigue Propensity Score of between 0 and 15. The ability of the Propensity Score to discriminate fatigued participants was estimated via receiver operating characteristic curve analysis.

Results

38% of participants reported significant fatigue during radiotherapy. Fatigue risk was predicted by elevated pre-treatment fatigue and anxiety, and diagnoses other than invasive ductal carcinoma (ductal carcinoma in-situ, invasive lobular and rarer carcinoma subtypes). The positive predictive value of the prognostic model was 80%. A Propensity Score threshold of ≥6 corresponded to a specificity of 90.3% and a sensitivity of 76.3%. The area under the receiver operating characteristic curve was 0.83 for the cross-validation sample.

Conclusions

Application of the Fatigue Propensity Score in the patient pathway can help direct fatigue management resources at those patients most likely to benefit.

Keywords: Fatigue, Breast cancer, Radiotherapy, Risk prediction.

Introduction

Patients rate fatigue as the most prevalent and severe untreated symptom during radiotherapy.1 and 2 As fatigue impacts adversely on all areas of normal functioning, a reduced quality of life is commonly reported.3 The effect of fatigue on patient's and primary caregiver's ability to work represents a cost borne by both individuals and society.2 These profound effects may be keenly felt by women with breast cancer who are often relatively young with busy lives.4

Approximately 30% of women who undergo radiotherapy for early breast cancer report sustained fatigue even years after completing active treatment.5 and 6 As acute fatigue is a strong risk factor for consequential chronic fatigue7 and 8 the optimal scenario would be early intervention in the treatment pathway; to prevent fatigue arising, or at least attenuate intensification.

Cancer-related fatigue (CRF) has the potential for amelioration through both pharmacological and non-pharmacological pathways.9 and 10 The management of treatment-related fatigue, however, remains the exception. This deficiency is due to a complex aetiology and uncertainty regarding which patients will experience significant fatigue.

The incidence of significant fatigue depends on the tool used to measure it. At least nine measures have been used in breast cancer populations,11 although not all have been adequately validated in this context.

A methodologically robust review of validated cancer-specific measures recommends the widely used Functional Assessment of Chronic Illness Therapy Fatigue Scale (FACIT-F) for research.12 Advantageous features of FACIT-F include brevity, excellent psychometric properties, an established threshold for the presence of fatigue and a quantified minimally clinically important difference in scores.13 and 14

This study reports on the development and validation of a prognostic tool that aims to predict women at a high risk of experiencing significant fatigue during radiotherapy – as measured by the FACIT-F. This Fatigue Propensity Score provides an objective base for the efficient targeting of fatigue interventions at those most likely to benefit.

Patients and methods

Patients

A local Research Ethics Committee granted the study ethical approval. The sample has been described previously.15 In brief, the population comprised women with stages Tis–T2N1 breast cancer referred to Velindre Cancer Centre for 40 Gy in 15 fractions over three weeks radiotherapy. Principal exclusion criteria were uncontrolled heart, lung, liver disease or thyroid dysfunction; pre-existing chronic autoimmune or inflammatory disease; previous radiotherapy exposure; locally advanced or metastatic disease; prior or concurrent use of systemic therapeutic agents.

A pragmatic sample size of 100 was based on a recruitment period of 12 months, and a projected recruitment rate of two-per-week. The rule of thumb of 10 (fatigue) cases per candidate independent variable was applied to minimise problems of ‘overfitting’ of the model.16 As the prevalence of fatigue was approximately 40%, this meant four independent variables.

Independent variables

The complex aetiology of CRF necessitated a theoretical framework to guide the selection of candidate variables. The sickness behaviour framework, as described by Dantzer and Kelley (2007), fulfilled this function.17 This framework seeks to explain the behavioural response to non-specific tissue injury through integration of psycho-neuro-immunological pathways. The framework principles were applied to the current oncological/radio-biological setting. Additional covariates of fatigue identified in the literature were also considered.

Observations included age, body mass index, smoking and hormone replacement therapy history, employment status, differential blood counts, concentration of the inflammatory mediator IL-6sR, histological diagnosis, histopathological grade, pathological stage grouping, disease laterality, volume of irradiated tissue, time elapsed from surgery to radiotherapy, and the travel time for daily radiotherapy.

Fatigue over the previous seven days was measured using the FACIT-F – at baseline two weeks before radiotherapy, at the end of the second and third weeks of radiotherapy, and four weeks after radiotherapy. FACIT-F has demonstrated test-retest reliability (r = 0.90), internal consistency (α = 0.93–0.95),13 sensitivity to change in haemoglobin and performance status18 and 19 and convergent validity against other scales. These properties lessened the value of including more than one measure of fatigue, which would complicate the model derivation and interpretation.

FACIT-F generates a raw score that is subtracted from 52 to render an inverted scale score between zero (maximum fatigue) and 52 (minimum fatigue). A scale score cut-off of 34 optimises classification of fatigued/not fatigued, as defined by the ICD-10 diagnostic criteria.20 and 21 Participants were classified as fatigued when the mean of scale scores at week two and three of radiotherapy was ≤34, or non-fatigued for >35.

A measure of anxiety and depression over the previous seven days was recorded using the Hospital Anxiety and Depression Scale (HADS).22 The scale has been validated in cancer patients23 and used to assess mood disorders in breast irradiation studies.24 and 25

The International Physical Activity Questionnaire (IPAQ) provided a total activity score over the previous seven days, plus low, moderate and high activity categorisation.26 The IPAQ has been used widely for population surveillance27 and in breast cancer studies.28

Derivation of the prognostic model and Fatigue Propensity Score

Characteristics of the two fatigue groups were examined with descriptive statistics. Categorical variables were transformed into multiple dichotomous variables before bivariate screening analyses to exclude variables not related to the outcome ‘fatigued or non-fatigued’ at the p < 0.25 level.29 The unadjusted odds ratio (OR) for statistically significant variables was estimated.

Multiple logistic regression analyses were undertaken for the retained variables. The dichotomised ‘fatigued/non-fatigued’ outcome was chosen as simple models are more likely to be used by clinicians.30 As the objective was prognostic, a (backwards-entry) stepwise algorithm that selects variables solely on statistical association was appropriate.31 The significance level criterion for exclusion, after the application of the likelihood ratio test, was p = 0.05. Adjusted ORs with 95% confidence intervals (CI) were estimated.

A risk weighting was applied to the final model variables, based on the underlying logistic regression equation coefficients. These risk factors were added together to render a Fatigue Propensity Score, with higher scores representing an elevated probability of fatigue.32 Finally, a score threshold was sought that best discriminated women at a high risk of fatigue.33

Validation of the model and Fatigue Propensity Score

The ability of the model to predict the outcome for individuals not in the study was assessed by a leave-one-out method.34 and 35 This cross-validation technique involves sequentially removing each individual case from the derivation sample and recalculating the regression coefficients. The probability of each of the 100 validation cases being fatigued could then be predicted from the derivation sample.

The receiver operating characteristic (ROC) curve is a standard statistical method used to assess the ability of a prediction score to discriminate individuals with an outcome from those without it. The area under the ROC curve provides a global measure of discriminatory accuracy.36 In the current context, the area under the ROC curve indicates the probability of a participant from the fatigued group having a higher score than a randomly chosen participant from the non-fatigued group.37 and 38 The area under the ROC curve for the derivation sample, the validation sample and the Fatigue Propensity Score were calculated and compared.35 Statistical analysis was performed using SPSS version 16 [SPSS Inc., Chicago IL., USA].

Results

Participant characteristics

130 study eligible women were approached to participate. 15 women were excluded on the basis of a full medical history; 15 eligible women declined to participate. Non-participants did not appear to be substantively different on important aspects: mean (SD) age and median (IQR) FACIT-F score were 57.1 (7.4) and 43 (37–46), respectively. All other women consented to their inclusion in the study.

Based on the a-priori criteria for fatigue, 38% of participants were classified as fatigued, with the remaining 62% classified as non-fatigued. The difference in median FACIT-F scores between the two groups was 13 before treatment, 23 at week three and 16 four weeks later (Fig. 1). The pre-radiotherapy characteristics for each group are presented in Table 1.

gr1

Fig. 1 Median longitudinal Functional Assessment of Chronic Illness Therapy Fatigue Scale (FACIT-F) score for fatigued and non-fatigued groups.

Table 1 Pre-radiotherapy participant characteristics of the fatigued and non-fatigued groups.

Characteristic Fatigued group (n = 38) Non-fatigued group (n = 62) Test significance p-value
Socio-demographics
Age (years) 55.9 (9.3) 59.0 (8.5) 0.1
Body mass index (Kg/m2) 29.2 (4.8) 27.6 (4.6) 0.1
HRT history     0.6
Never 25 (65.8%) 45 (72.6%)  
Previous 13(24.2%) 17 (27.4%)  
Smoking pack-yearsa 12.6 (16.4) 6.2 (9.7) 0.09
Employment status     0.6
Retired/Housewife 24 (63.2%) 38 (61.3%)  
Working 14 (36.8) 24 (38.7%)  
IL6-sR concentration (ng/dL) 43.5 (11.6) 39.8 (11.5) 0.1
Haematological
Red blood count (10*12/L) 4.41 (0.32) 4.42 (0.34) 0.2
Haemoglobin (g/dL) 13.4 (1.1) 13.7 (0.9) 0.2
Haematocrit 0.40 (0.03) 0.41 (0.03) 0.3
Platelets (10*9/L) 264.2 (60.5) 270.0 (69.3) 0.7
White blood count (10*9/L)a 6.3 (2.2) 6.8 (1.9) 0.2
Neutrophils (10*9/L)a 3.6 (1.3) 3.9 (1.5) 0.2
Lymphocytes (10*9/L)a 2.0 (0.7) 2.1 (0.8) 0.5
Monocytes (10*9/L)a 0.52 (0.20) 0.46 (0.17) 0.7
Eosinophils (10*9/L)a 0.19 (0.16) 0.19 (0.15) 0.8
Basophils (10*9/L)a 0.03 (0.02) 0.03 (0.02) 0.2
Disease-related
Histological diagnosis     0.05
Ductal carcinoma in-situ 6 (15.8%) 7 (11.3%)  
Invasive ductal carcinoma 18 (47.4%) 43 (69.4%)  
Invasive lobular carcinoma 5 (13.2%) 7 (11.3%)  
Rarer subtypes 9 (23.7%) 5 (8.1%)  
Histopathological grade     0.5
Grade 1 14 (36.8%) 17 (27.4%)  
Grade 2 17 (44.7%) 34 (54.8%)  
Grade 3 7 (18.4%) 11 (17.7%)  
TNM group stage     0.3
0 7 (18.4%) 6 (9.7%)  
I 25 (65.8%) 45 (72.6%)  
IIA 6 (15.8%) 11 (17.7%)  
Laterality     0.9
Right 20 (52.6%) 32 (51.6%)  
Left 18 (47.4%) 30 (48.4%)  
Time from surgery to RT (days) 61.5 (12.2) 61.0 (16.7) 0.2
Travel time for RT (mins) 36.7 (19.1) 39.8 (34.3) 0.9
Self-reported questionnaires
FACIT-Fa 47.5 (6) 34.5 (15.25) <0.0005
HADS anxietya 8 (7.5) 4.3 (2.8) p = 0.001
HADS depressiona 4 (4.5) 1.9 (2.1) p < 0.001
IPAQ total (MET-min/wk)a 1759 (2742) 1752 (2166) p = 0.9
Low activity 16 (42.1) 21 (33.9)  
Medium activity 17 (44.7) 23 (37.1)  
High activity 5 (13.2) 18 (29.0)  

a Non-normally distributed continuous variables are medians (inter-quartile range).

Normally distributed continuous variables are means (standard deviation).

Categorical variables are numbers (and percentages) of patients in that group.

HRT = hormone replacement therapy; RT = radiotherapy; FACIT-F = Functional Assessment of Chronic Illness Therapy Fatigue Scale; HADS = Hospital Anxiety and Depression Scale; IPAQ = International Physical Activity Questionnaire.

Model derivation

Bivariate regression analysis was conducted to identify baseline variables that contribute to variance in fatigue status (p < 0.25). The Wald statistic was used, which indicates whether the logistic regression coefficient for a risk factor is significantly different from zero. The magnitude and direction of the relationship between the retained variables and the risk of fatigue was quantified as an unadjusted odds ratio estimate (Table 2).

Table 2 The likelihood of fatigue occurring based on unadjusted risk factors.

Risk factor OR 95% CI Wald p value
Age 0.96 0.91–1.01 0.09
BMI 1.07 0.98–1.17 0.12
Smoking pack years 1.04 1.01–1.07 0.02
Haemoglobin 0.73 0.48–1.11 0.14
Menopausal status (pre/peri versus post) 0.46 0.18–1.16 0.10
Diagnosis (invasive ductal ca. versus others) 0.55 0.20–1.40 0.19
Raw FACIT-F fatigue score 0.85 0.80–0.91 <0.00001
HADS anxiety score 1.26 1.11–1.43 <0.0001
HADS depression score 1.49 1.23–1.80 <0.0001
IPAQ physical activity (high category versus others) 0.37 0.13–1.10 0.07

The retained variables were entered into a multiple logistic regression analysis with a threshold for retention of p ≤ 0.05 for the likelihood ratio statistic. The final fatigue probability model comprised three pre-treatment risk factors: raw FACIT-F fatigue score, HADS anxiety score and diagnoses other than invasive ductal carcinoma. The adjusted odds ratios (AOR) after controlling for contributions from the other variables in the model are presented in Table 3.

Table 3 The final fatigue probability model comprising adjusted risk factors.

Risk factor AOR 95% CI Wald p value
Raw FACIT-F fatigue score 1.15 1.07–1.23 <0.0001
HADS anxiety score 1.18 1.02–1.37 0.03
Diagnosis (‘other diagnoses’ versus invasive ductal ca.) 3.00 1.03–8.80 0.04

As FACIT-F is a continuous variable the interpretation of the associated OR is that for every additional point increase in raw pre-treatment FACIT-F score participants were 1.15 times (15%) more likely to be classified as fatigued during treatment. Elevated pre-treatment anxiety and histological diagnoses other than invasive ductal carcinoma were also statistically significant predictors of fatigue, after controlling for other variables.

The final regression model had utility in classifying participants to the fatigued/non-fatigued groups, χ2 (3, N = 100) = 45.6, p < 0.0001. The model correctly classified 82% of all patients. The model sensitivity was 70.1%, the specificity was 90.3% and the positive predictive value was 80%.

Fatigue Propensity Score

The relative contributions of the final model variables were maintained in the application of a risk weighting. A Fatigue Propensity Score of up to 15 was calculated for each participant, using the formula:

FORMULA:

stripin: si1.gif

To illustrate, a woman reporting the mean cohort baseline raw FACIT-F score of 11, HADS anxiety score of 5.5 and a diagnosis of invasive ductal carcinoma has a Propensity Score of 5.3 and approximately 30% predicted probability of being in the fatigued group. A subject with one SD difference in baseline values, in the poorer direction, i.e. FACIT-F of 21, HADS anxiety of 9.4 and the same diagnosis, has a Propensity Score of 7.2 and a 70% predicted probability of being classified as fatigued. The probability of fatigue as a function of Propensity Score is illustrated in Fig. 2.

gr2

Fig. 2 Fatigue Propensity Score as a function of the probability of fatigue.

A Propensity Score-threshold was sought that discriminated women at a high risk of fatigue. A cut-off score of ≥6 corresponded to a specificity of 90.3% and sensitivity of 76.3%.

Model validation

The model appeared to be stable during the leave-one-out cross-validation, with minimal decrement in predictive ability when applied to the validation dataset compared to the derivation data set: percentage correctly predicted 81.9 (95% CI 81.8–82.1) versus 82.1, sensitivity 68.4 (68.3–68.6) versus 70.1 and specificity 90.3 (90.2–90.4) versus 90.3, respectively.

The discrimination ability of the derivation dataset, Fatigue Propensity Score and validation dataset was assessed by comparison of the area under the ROC curve. The respective results were 0.87 (95% CI 0.80–0.94), 0.86 (0.78–0.94) and 0.83 (0.76–0.96). An area under the curve of 0.8 is considered to represent good discrimination ability, whilst 0.5 is equivalent to a tossing a coin to predict fatigue status.38

Discussion

The final model, developed to predict the probability of fatigue in women undergoing radiotherapy for breast cancer, comprised three pre-treatment risk factors: FACIT-F fatigue score, HADS anxiety score and a diagnosis other than invasive ductal carcinoma.

Pre-treatment fatigue made the largest individual contribution to the predictive ability of the model, with an estimated large effect size R = 0.45. For every point increase in raw baseline fatigue score the likelihood of the participant being in the fatigued group increased by between seven and 23%. The actual value is likely to be nearer the point estimate of 15%. Fig. 1 indicates that for a significant minority of patients determinants of fatigue precipitate a pre-treatment level of fatigue that is subsequently exacerbated during radiotherapy. The nature of these precipitating factors is outside the scope of this study, but may be related to primary surgery, individual's inflammatory profile,39 personality traits40 and psychological response to diagnosis and prospective treatment.41

The discrete nature of the fatigue groups suggests that pre-treatment fatigue alone is a useful predictor of subsequent fatigue. The joint inclusion of the other two variables increases the ability to discriminate an individual case of fatigue by approximately 12%. Relying on a single fatigue score (albeit referring to the previous seven days) may be susceptible to random fluctuations and regression to the mean problems, especially when applied to an external sample.

Each point increase in baseline HADS anxiety is associated with an increased likelihood of being in the fatigued group of 18%. The data was consistent with an increased likelihood of up to 37%. Substantial evidence reports a correlation between anxiety and CRF.42 The causal nature of this relationship remains uncertain, however, psychological, immunological and neuro-endocrine theory collectively support the concept that anxiety can trigger behavioural symptoms such as fatigue.43, 44, and 45 We also know that, as a group, younger women with early breast cancer report relatively elevated levels of anxiety.46

Women with diagnoses other than invasive ductal carcinoma, in this study, ductal carcinoma in-situ, invasive lobular and mucinous, medullary and tubular carcinoma, were three times more likely to be in the fatigued group. Participants with ductal carcinoma in-situ recorded both an elevated median fatigue score at baseline (38) compared to all other diagnoses (46), and a greater intensity of anxiety. The lower psychological mood in this subgroup, which was not related to lesser age, may be related to inaccurate perceptions of the risk of this pre-cancerous pathology and uncertainty regarding the disease natural history, and therefore the necessity of treatment.47

Contrary to a previous report48 the current study found little evidence that haematological variables are prognostic for subsequent fatigue. Physical activity was also not a useful predictor of fatigue, although a higher proportion of women in the non-fatigued group met the classification for high physical activity.

Overall, the model classified 82% of the study participants to the correct fatigue group. The proportion of participants predicted to be fatigued who were classified correctly was 80%.

A common weakness of statistical models is that they fit the derivation dataset well, but suffer a decrement in predictive ability when applied to a secondary dataset. The current model maintained good stability during the cross validation procedure and the ROC curve analysis. However, the main limitation of this study remains the lack of an external validation sample. A full validation requires a secondary sample drawn at random.

A second limitation is that the model is derived from a homogenous group of patients with stage groups 0–IIA disease. Only three participants received a (10 Gy) radiation boost and only two participants received regional irradiation, to the ipsilateral supraclavicular fossa. Generalisability of results to patients who have received prior adjuvant systemic treatments is uncertain. The model validation should therefore be extended to assess performance in patients who have received adjuvant chemotherapy.

Thirdly, there may be other useful prognostic factors that were not observed, and uncontrolled confounders that may introduce bias. For example, participants reported taking statins, anti-hypertensive, diuretic and gastric medication, which could modulate fatigue.

The development of the Fatigue Propensity Score enables clinicians to discriminate patients at a high risk of fatigue from those at a lower risk. A score of six is the proposed threshold to classify high-risk patients. This threshold corresponds to a ‘highly’ specific tool (90.3%) with 80% of the cohort with a score of six or greater subsequently being in the fatigued group. Modulating this threshold varies the trade-off between sensitivity and specificity, in accordance with clinical judgement or resource constraints.37

The early identification of fatigue risk shifts the emphasis from a reaction to existing symptoms to proactive fatigue prevention. The Fatigue Propensity Score provides an objective criterion to target resources at patients likely to benefit. Even the most active fatigue intervention will be unimpressive for the 62% of women who experienced a clinically insignificant intensity of fatigue before, during or after radiotherapy. These patients can be reassured that fatigue symptoms are likely to be mild, transitory and manageable by simple lifestyle adaptations. The ability to inform patients of the likelihood of iatrogenic toxicities is set to become an increasingly important aspect of the patient consent process, as the range of breast cancer treatments with equivalent actuarial rates expands.

Fatigue prediction can also be included as a covariate/eligibility criterion when testing experimental fatigue interventions (which the results suggest should incorporate anxiety reduction strategies). This enables discrimination between interventions that reduce fatigue and those that prevent fatigue.

The Fatigue Propensity Score has potential to facilitate the attenuation of acute fatigue and prevent chronic fatigue and related behavioural problems arising, with positive implications for adaptation to life as a cancer survivor and improved quality of life.

Study support

The data presented here was derived from a PhD study that was funded by a grant from the Research Capacity Building Collaboration (RCBC) Wales.

The College of Radiographers Industrial Partnership Scheme Research Awards funded the equipment and laboratory consumables to conduct immunological assays.

Conflict of interest statement

None.

Acknowledgements

The study was funded by the Research Capacity Building Collaboration (RCBC) Wales and the College of Radiographers Industrial Partnership Scheme Research Awards. We would like to thank the Cardiff University staff at the Cancer Research Wales laboratories for assistance with the cytokine assays. The funders had no role the study design, in the collection, analysis and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication.

References

  • 1 J.T. Hickok, G.R. Morrow, J.A. Roscoe, K. Mustian, P. Okunieff. Occurrence, severity, and longitudinal course of twelve common symptoms in 1129 consecutive patients during radiotherapy for cancer. J Pain Symptom Manage. 2005;30:433-442 Crossref.
  • 2 G.A. Curt, W. Breitbart, D. Cella, J.E. Groopman, S.J. Horning, L.M. Itri, et al. Impact of cancer-related fatigue on the lives of patients: new findings from the Fatigue Coalition. Oncologist. 2000;5:353-360 Crossref.
  • 3 P.C. Dagnelie, M.C.G. Pijls-Johannesma, P. Lambin, S. Beijar, D. De Ruysscher, G.I.J.M. Kempen. Impact of fatigue on overall quality of life in lung and breast cancer patients selected for high-dose radiotherapy. Ann Oncol. 2007;18:940-944 Crossref.
  • 4 D. Lowdermilk, B.B. Germino. Helping women and their families cope with the impact of gynecologic cancer. J Obstet Gynecol Neonatal Nurs. 2000;29:653-660 Crossref.
  • 5 S. Alexander, O. Minton, P. Andrews, P. Stone. A comparison of the characteristics of disease-free breast cancer survivors with or without cancer-related fatigue syndrome. Eur J Cancer. 2009;45:384-392 Crossref.
  • 6 J.E. Bower, P.A. Ganz, K.A. Desmond, C. Bernaards, J.H. Rowland, B.E. Meyerowitz, et al. Fatigue in long-term breast carcinoma survivors. Cancer. 2006;106:751-758 Crossref.
  • 7 H. Geinitz, F.B. Zimmermann, R. Thamm, M. Keller, R. Busch, M. Molls. Fatigue in patients with adjuvant radiation therapy for breast cancer: long-term follow-up. J Cancer Res Clin Oncol. 2004;130:327-333 Crossref.
  • 8 E.M. Smets, V.R. Visser, F.M. Willems-Groot, B. Garssen, A.L. Schuster-Uitterhoeve, J.C. de Haes. Fatigue and radiotherapy: (B) experience in patients 9 months following treatment. Brit J Cancer. 1998;78:907-912
  • 9 J.K. Carroll, S. Kohli, K.M. Mustian, J.A. Roscoe, G.R. Morrow. Pharmacologic treatment of cancer-related fatigue. Oncologist. 2007;12:43-51
  • 10 K.M. Mustian, G.R. Morrow, J.K. Carroll, C.D. Figueroa-Moseley, P. Jean-Pierre, G.C. Williams. Integrative nonpharmacologic behavioural interventions for the management of cancer-related fatigue. Oncologist. 2007;12:52-67
  • 11 G. Prue, J. Rankin, J. Allen, J. Gracey, F. Cramp. Cancer-related fatigue: a critical appraisal. Eur J Cancer. 2006;42:846-863 Crossref.
  • 12 O. Minton, P. Stone. A systematic review of the scales used for the management of cancer-related fatigue (CRF). Ann Oncol. 2009;20:17-25
  • 13 D. Cella. Manual of the Functional Assessment of Chronic Illness Therapy (FACIT) measurement system. (Evanston Northwestern Healthcare and Northwestern University, Evanston, IL, 1997)
  • 14 D.F. Cella, D.T. Eton, S.J. Lai, A. Peterman, D.E. Merkel. Combining anchor and distribution-based methods to derive minimal clinically important differences on the Functional Assessment of Cancer Therapy (FACT) anemia and fatigue scales. J Pain Symptom Manage. 2002;24:547-561 Crossref.
  • 15 N. Courtier, T. Gambling, S. Enright, P. Barrett-Lee, J. Abrahams, M.D. Mason. Psychological and immunological characteristics of fatigued women undergoing radiotherapy for early-stage breast cancer. Support Care Cancer. 2012;10.1007/s00520-012-1508-6
  • 16 F.E. Harrell, K.L. Lee, R.M. Califf, D.B. Pryor, R.A. Rosati. Regression modelling strategies for improved prognostic prediction. Stat Med. 1984;3:143-152 Crossref.
  • 17 R. Dantzer, K.W. Kelley. Twenty years of research on cytokine-induced sickness behaviour. Brain Behav Immun. 2007;21:153-160 Crossref.
  • 18 D. Cella, S. Lai, C.H. Chang, A. Peterman, M. Slavin. Fatigue in cancer patients compared with fatigue in the general United States population. Cancer. 2002;94:528-538 Crossref.
  • 19 E. Berndt, J. Kallich, A. McDermott, X. Xu, H. Lee, J. Glaspy. Reductions in anaemia and fatigue are associated with improvements in productivity in cancer patients receiving chemotherapy. Pharmacoeconomics. 2005;23:505-514 Crossref.
  • 20 M.A. Andrykowski, J.E. Schmidt, J.M. Salsman, A.O. Beacham, P.B. Jacobsen. Use of a case definition approach to identify cancer-related fatigue in women undergoing adjuvant therapy for breast cancer. J Clin Oncol. 2005;23:6613-6622 Crossref.
  • 21 S. Van Belle, R. Paridaens, G. Evers, J. Kerger, D. Bron, J. Foubert, et al. Comparison of proposed diagnostic criteria with FACT-F and VAS for cancer related fatigue: proposal for use as a screening tool. Support Care Cancer. 2005;13:246-254
  • 22 A.S. Zigmond, R.P. Snaith. The hospital anxiety and depression scale. Acta Psychiatr Scand. 1983;67:361-370 Crossref.
  • 23 S. Moorey, S. Greer, M. Watson, C. Gorman, L. Rowden, R. Tunmore, et al. The factor structure and factor stability of the hospital anxiety and depression scale in patients with cancer. Brit J Psychol. 1990;158:255-259
  • 24 H. Geinitz, F.B. Zimmermann, P. Stoll, R. Thamm, W. Kaffenberger, K. Ansorg, et al. Fatigue, serum cytokine levels, and blood cell counts during radiotherapy of patients with breast cancer. Int J Radiat Oncol Biol Phys. 2001;51:691-698 Crossref.
  • 25 R. Maraste, L. Brandt, H. Olsson, B. Ryde-Brandt. Anxiety and depression in breast cancer patients at start of adjuvant radiotherapy: relations to age and type of surgery. Acta Oncol. 1992;31:641-643 Crossref.
  • 26 IPAQ. Guidelines for data processing and analysis of the International Physical Activity Questionnaire (IPAQ) – short and long forms. (, November 2005) http://www.ipaq.ki.se/dloads/IPAQ%20LS%20Scoring%20Protocols_Nov05.pdf Available from:
  • 27 C.L. Craig, A.L. Marshall, M. Sjöström, A.E. Bauman, M.L. Booth, B.E. Ainsworth, et al. International Physical Activity Questionnaire: 12 country reliability and validity. Med Sci Sports Exerc. 2003;35:1381-1395 Crossref.
  • 28 M. Johnson-Kozlow, J.F. Sallis, E.A. Gilpin, C.L. Rock, J.P. Pierce. Comparative validation of the IPAQ and the 7-day PAR among women diagnosed with breast cancer. Int J Behav Nutr Phys Act. 2006;3:7 Crossref.
  • 29 S.L. George. Identification and assessment of prognostic factors. Semin Oncol. 1988;15:462-471
  • 30 L.E. Braitman, F. Davidoff. Predicting clinical states in individual patients. Ann Intern Med. 1996;125:406-412 Crossref.
  • 31 M.H. Katz. Multivariable analysis: a practical guide for clinicians. 2nd ed. (Cambridge University Press, New York, 2006) p. 114
  • 32 P. Barrett-Lee, C. Bokemeyer, P. Gascon, J.W.R. Nortier, M. Schneider, D. Schrijvers, et al. Management of cancer-related anemia in patients with breast or gynecologic cancer: new insights based on results from the European Cancer Anemia Survey. Oncologist. 2005;10:743-757 Crossref.
  • 33 G. Dranitsaris, M. Clemons, S. Verma, C. Lau, M. Vincent. Chemotherapy-induced anaemia during adjuvant treatment for breast cancer: development of a prediction model. Lancet Oncol. 2005;6:856-863 Crossref.
  • 34 M. Stone. Cross-validatory choice and assessment of statistical predictions. J Roy Stat Soc M Met. 1974;36:111-147
  • 35 D. Bautista, E. Arana, L. Martí-Bonmatí, R. Paredes. Validation of logistic regression models in small samples: application to calvarial lesions diagnosis. J Clin Epidemiol. 1999;52:237-241 Crossref.
  • 36 D.G. Altman, J.M. Bland. Statistics notes: diagnostic tests 3: receiver operating characteristic plots. BMJ. 1994;309:188 Crossref.
  • 37 S.B. Cantor, M.W. Kattan. Determining the area under the ROC curve for a binary test. Med Decis Making. 2000;20:468-470 Crossref.
  • 38 J.A. Hanley, B.J. McNeil. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143:29-36
  • 39 J.E. Bower. Behavioural symptoms in patients with breast cancer and survivors. J Clinl Oncol. 2008;26:768-777 Crossref.
  • 40 J. De Vries, A.F. Van der Steeg, J.A. Roukema. Trait anxiety determines depressive symptoms and fatigue in women with an abnormality in the breast. Brit J Health Psychol. 2009;14:143-157
  • 41 B. Bennett, D. Goldstein, A. Lloyd, T. Davenport, I. Hickie. Fatigue and psychological distress-exploring the relationship in women treated for breast cancer. Eur J Cancer. 2004;40:1689-1695 Crossref.
  • 42 L.F. Brown, K. Kroenke. Cancer-related fatigue and its associations with depression and anxiety: a systematic review. Psychosomatics. 2009;50:440-447
  • 43 A.H. Miller, S. Ancoli-Israel, J.E. Bower, L. Capuron, M.R. Irwin. Neuroendocrine-immune mechanisms of behavioural comorbidities in patients with cancer. J Clin Oncol. 2008;26:971-982 Crossref.
  • 44 B.A. McGregor, M.H. Antoni. Psychological intervention and health outcomes among women treated for breast cancer: a review of stress pathways and biological mediators. Brain Behav Immun. 2009;23:159-166 Crossref.
  • 45 J.E. Bower, P.A. Ganz, N. Aziz. Altered cortisol response to psychologic stress in breast cancer survivors. Psychosom Med. 2005;67:277-280 Crossref.
  • 46 P. Hopwood, J. Haviland, J. Mills, G. Sumo, J. Bliss. The impact of age and clinical factors on quality of life in early breast cancer: an analysis of 2208 women recruited to the UK START Trial (Standardisation of Breast Radiotherapy Trial). Breast. 2007;16:241-251 Crossref.
  • 47 A. Partridge, K. Adloff, E. Blood, C. Dees, C. Kaelin, M. Golshan, et al. Risk perceptions and psychosocial outcomes of women with ductal carcinoma in situ: longitudinal results from a cohort study. J Natl Cancer Inst. 2008;100:243-251 Crossref.
  • 48 C. Wratten, J. Kilmurray, S. Nash, M. Seldon, C.S. Hamilton, P.C. O'Brien, et al. Fatigue during breast radiotherapy and its relationship to biological factors. Int J Radiat Oncol Biol Phys. 2004;59:160-167 Crossref.

Footnotes

a School of Healthcare Studies, Cardiff University, Cardiff, UK

b Velindre Cancer Centre, Velindre NHS Trust, Cardiff, UK

Corresponding author. Department of Radiography, School of Healthcare Studies, Cardiff University, Heath Park, Cardiff CF14 4XN, UK. Tel.: +44 (0) 29 20 687566; fax: +44 (0) 29 20 687694.


Search this site

Stay up-to-date with our monthly e-alert

If you want to regularly receive information on what is happening in Quality of Life in Oncology research sign up to our e-alert.

Subscribe »

QOL (Quality of Life) newsletter e-alert

NEW! Free access to the digital version of a new publication in Cancer Supportive Care


Cancer cachexia: mechanisms and progress in treatment

Authors: Egidio Del Fabbro, Kenneth Fearon, Florian Strasser

This book was supported by an educational grant from Helsinn Healthcare SA.

Featured videos

Quality of Life promotional video

Made possible by an educational grant from Helsinn

Helsinn does not have any influence on the content and all items are subject to independent peer and editorial review

Society Partners

European Cancer Organisation Logo

Share