ORIGINAL ARTICLES LIVER, PANCREAS, AND BILIARY TRACT A

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6: ORIGINAL ARTICLES LIVER, PANCREAS, AND BILIARY TRACT A Predictive Model for Fatigue and Its Etiologic Associations in Primary Biliary Cirrhosis JULIA L. NEWTON,*, JESSIE PAIRMAN, KATHRYN SUTCLIFFE, KATHARINE WILTON, and DAVID E. J. JONES* *Liver Research Group, Institute of Cellular Medicine, Newcastle University, and Cardiovascular Investigation Unit, Royal Victoria Infirmary, Newcastle-upon-Tyne, United Kingdom Background & Aims: Excessive day-time somnolence and autonomic dysfunction are biological processes prevalent in Primary Biliary Cirrhosis (PBC) that associate with fatigue. Here we explore how these biological associates inter-relate, and their cumulative impact upon typical clinical cohorts. Methods: A predictive model for daytime hypersomnolence (Epworth Sleepiness Scale (ESS)) and autonomic dysfunction (Orthostatic Grading Scale (OGS)) was developed in a derivation cohort (n 124) and subsequently validated in a second cohort (n 114). Subjects also completed the disease specific quality of life tool, the PBC-40. Results: A composite predictive criterion (presence of either ESS >10 or OGS >4) for the presence of fatigue in PBC patients had a sensitivity of 0.71 (95% confidence intervals ) and specificity 0.8 ( ) (positive predictive value (PV); 0.84 ( ), negative PV; 0.66 ( ) for moderate or severe fatigue). Ninety-seven percent of severely fatigued patients (0% of non-fatigued) met the aetiology predictive criterion ( , P<.0001). Expression of both significant daytime somnolence and autonomic dysfunction was not associated with more severe fatigue, suggesting that there is a threshold effect for fatigue in PBC. When applied to a second independent cohort, the composite criterion retained strongly significant predictive value for fatigue. Conclusions: A significant proportion of fatigue in PBC associates with one or both of autonomic dysfunction (OGS >4) and sleep disturbance (ESS >10). Those meeting both ESS and OGS criteria were not more severe fatigued than those meeting the diagnostic criterion for either OGS or ESS alone. A threshold effect for fatigue has implications for potential therapeutic interventions. Primary biliary cirrhosis (PBC) is a chronic cholestatic liver disease that affects up to 1 in 700 women older than the age of 45 in the United Kingdom. 1 For many years the principle clinical goal in PBC has been to slow the progression to endstage liver disease with its associated morbidity and mortality. Although this remains an important clinical goal (PBC is associated with significant mortality within its affected patient group including a significant component resulting directly from end-stage liver disease), it is becoming clear that PBC patients experience additional clinical problems, and that there are new clinical challenges to be faced by physicians who manage PBC patients. 2 PBC patients experience significant impairment of their quality of life for reasons that are largely independent of the severity of their underlying liver disease. 3 5 Although the impact of itch in the experience of PBC patients has been recognized for a long time, only more recently has it become appreciated that a significant proportion of PBC patients also experience a symptom complex consisting of fatigue, symptoms suggestive of cognitive dysfunction, and social and emotional dysfunction. 6 Given the apparent impact of this fatigue-based symptom complex in PBC patients, significant recent effort has gone into understanding the mechanisms that might underpin it. Two significant biological processes have been identified recently that appear to be prevalent in PBC patients and to be associated with the fatigue symptom complex. The strongest association is with abnormality in sleep patterns with, in particular, excessive daytime somnolence. 3,7,8 Self-report of daytime somnolence in PBC patients is frequent and the degree of fatigue experienced by patients correlates with perception of degree of daytime sleep. 7 Moreover, objective sleep testing suggests a move from nighttime to daytime sleep, with a significant association between actual minutes of daytime sleep and fatigue severity. The presence of significant sleep abnormality in PBC patients has been postulated to reflect the presence of a chronic secondary central nervous system process occurring as a result of chronic inflammation and/or cholestasis. The second biological process seemingly associated with fatigue in PBC patients is autonomic dysfunction. Objective clinical features suggestive of autonomic dysfunction (including abnormality in heart rate variability, in blood pressure regulation in the context of the Valsalva maneuver, and after tilt testing) is frequent in PBC patients 9 and the degree of abnormality appears to be associated with severity of fatigue Population studies using validated symptom assessment tools suggest that symptoms Abbreviations used in this paper: CI, confidence interval; ESS, Epworth Sleepiness Scale; OGS, Orthostatic Grading Scale by the AGA Institute /08/$34.00 doi: /j.cgh

2 February 2008 FATIGUE MECHANISMS IN PBC 229 Table 1. Clinical Features of the Independent Development (NE1-25 Geographically Based Cohort) and Validation PBC Cohorts (Clinic-Based Cohort) Development cohort Validation cohort n Age, y Bilirubin level, mol/l Albumin level, g/l Alkaline phosphatase level, IU/L Disease stage at last biopsy (%) 29 (23)/29 (23)/66 (54) 60 (53)/32 (28)/22 (19) No biopsy/cirrhotic/precirrhotic NOTE. Values shown are mean SD unless stated otherwise. typical of autonomic dysfunction are frequent in the PBC population, and are themselves associated with fatigue. 10 The cause of autonomic dysfunction in PBC is likely to be complex with at least some contribution coming from the hemodynamic change associated with progressive liver disease. The presence, however, of significant autonomic dysfunction in patients with very early stage disease is suggestive of an additional mechanism postulated, again, to result from brain stem changes secondary to inflammatory cholestasis. The identification of sleep regulation abnormality and autonomic dysfunction as strong biological associates of fatigue in PBC is potentially of real clinical importance because both processes may be amenable to therapeutic intervention. If these processes are related causally to fatigue in PBC such interventions might be expected to be of benefit in terms of fatigue reduction. Early observations already have suggested that the anti-daytime somnolence drug modafinil has some merit in the treatment of fatigue in PBC, 8,13 whereas simple interventions aimed at ameliorating the effects of autonomic dysfunction, such as withdrawal of inappropriate antihypertensive agents, have shown similar early promise. 10 However, the logical exploration of such interventions requires a clearer understanding of how the identified biological associates of fatigue interact, and how they impact on typical clinical cohorts. In this study we set out to develop a clinical decision making model for the identification of significant sleep abnormality and autonomic dysfunction in PBC patients to establish a clinical predictive framework to allow the rational application of future therapeutic interventions. Patients and Methods Study Design This was a 2-phase study. A predictive model for sleep and autonomic dysfunction associated with fatigue in PBC was developed initially in a derivation cohort and subsequently was validated in an independent second cohort. The study was approved by our local research ethics committee. All subjects provided written permission for the use of clinical data and the local research ethics committee considered the return of questionnaires as implied consent. Study Cohorts The derivation cohort consisted of prevalent patients within the geographic postal code area NE1-25 at the study point. The derivation and validation of the NE1-25 cohort have been described extensively elsewhere. 14 The benefit of this clinical cohort is that it is representative of the whole PBC population and is not subject to clinic attendance bias (eg, as might result from particularly symptomatic patients attending a clinic more frequently). The validation cohort consisted of a group of patients attending a specialist PBC clinic but who were otherwise unselected with regard to symptomatology. Cohorts 1 and 2 were fully independent. The clinical details of patients in these clinical cohorts are outlined in Table 1. Symptom Assessment Tools PBC-40. Health-related quality of life and the symptoms that contribute to its impairment in PBC were assessed in the PBC study cohorts using the PBC-40, a fully validated PBC-specific, multidomain, quality-of-life measure. 15 The PBC-40 contains 40 questions in 5 domains (Fatigue, Itch, Cognitive, Social and Emotional, and Other Symptoms a domain relating to a number of PBC-related symptoms that did not map to the other domains). A recent study defined ranges of severity for the symptom domains contained in the PBC By using these clinically meaningful cut-off values applied to the scores from the PBC-40 Fatigue domain, no fatigue was a score of 11 or less, mild was a score of 12 to 28, moderate was a score of 29 to 39, and severe was a score of 40 or greater. Orthostatic Grading Scale. To determine the degree of orthostatic intolerance as an indicator of autonomic dysfunction the subjects completed the Orthostatic Grading Scale (OGS). This is a fully validated, self-report, assessment tool for the symptoms of orthostatic intolerance caused by orthostatic hypotension (eg, severity, frequency, and interference with daily activities) that consists of 5 items, each graded on a scale of 0 to Adding the scores for the individual items creates a total score. Studies have shown that scores from the OGS correlate with conventional tests of the integrity of the autonomic nervous system. A score greater than 9 is considered to be consistent with a formal diagnosis of orthostatic hypotension. A score of 4 or greater and less than 9 was regarded as being indicative of moderate orthostatic hypotension. 16 Epworth Sleepiness Scale. In view of the recently identified associations between excessive daytime sleep and fatigue in PBC all subjects completed the Epworth Sleepiness Scale (ESS; possible score range, 0 24). This fully validated tool assesses daytime hypersomnolence, a score of 10 or more being indicative of significant daytime hypersomnolence. 17 Statistical Analysis Analysis was performed using Graphpad Prism Software (Graphpad Prism Software Inc, San Diego, CA). All vari-

3 230 NEWTON ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 2 ables were parametric. Comparisons between groups were made using unpaired t tests and chi-squared tests where appropriate. A statistically significant result was taken at the conventional significance level of 5%. Results The derivation cohort consisted of 124 patients with PBC who were residing within the geographic area defined by post codes NE1-25 at the study point. The clinical details for this patient group are given in Table 1. As previously shown, the ESS and OGS had significant utility when used independently for the prediction of the presence of significant fatigue (defined as the presence of moderate or severe fatigue using the previously described diagnostic cut-off values for the PBC in this cohort; Figure 1). Interestingly, medication with ursodeoxycholic acid was not associated with significant differences in ESS (no ursodeoxycholic acid, 7.3 [score 5] vs ursodeoxycholic acid, 7.2 [score 4.6]; P.9) or OGS scores (3.3 [score 3.5] vs 3.0 [score 3.10]; P.6). The previously defined criteria for significant daytime somnolence as assessed using the ESS ( 10) and significant vasomotor autonomic dysfunction using OGS ( 4) had optimal value for the prediction of the presence of moderate or severe fatigue in the derivation cohort (Figure 2A and B). This observation again confirms our previous findings. Although a significant association was seen between OGS and ESS scores in individual patients, the correlation was relatively weak (r , Figure 2C). This observation suggested a significant degree of independence between individual ESS and OGS scores. In light of this finding we explored the predictive value of a composite predictive criterion (presence of either ESS 10 or OGS 4) for the presence of fatigue in PBC patients. This composite predictive criterion had a sensitivity of 0.71 (95% confidence interval [CI], ) and specificity 0.8 (95% CI, ) with a positive predictive value of 0.84 (95% CI, ) and a negative predictive value of 0.66 (95% CI, ) for moderate or severe fatigue (Figure 2D). When this predictive criterion was applied across the full range of fatigue severity using the previously defined cut-off values extremely strong, step-wise predictive utility was seen (Figure 3). The etiology Figure 1. Predictive value of ESS (solid line) and OGS (dashed line) scores for the presence of moderate or severe fatigue in PBC patients in the derivation cohort 1 assessed by receiver operating characteristic analysis. ESS area under the curve, 0.82 (95% CI, ); OGS area under the curve, 0.78 (95% CI, ); both P predictive criterion was met in 97% of severely fatigued patients and in 0% of patients showing no fatigue ( 2, 49.6; P.0001). Unsurprisingly, given the predictive value of the criterion, PBC-40 Fatigue domain scores were substantially higher in patients screening positive than in patients screening negative (Figure 4). A significant minority of patients experiencing, in particular, moderate fatigue, however, were negative for the predictive criterion, suggesting the presence of at least one further etiologic factor for fatigue in PBC. Interestingly, when we explored the extent to which the associations between fatigue and significant daytime somnolence and autonomic symptomatology as assessed by ESS and OGS were additive it became clear that the presence of either of these associated features was sufficient to generate an association with fatigue (Figure 5). Strikingly, however, the possession of both of these associated features was not related to any apparent increase in fatigue severity. These findings are compatible with there being a threshold effect for the generation of fatigue. The studies performed in the derivation cohort established the fatigue etiology predictive criterion of ESS 10 and/or OGS 4 as a robust predictor for the presence of fatigue. To confirm its validity we went on to apply it to a second independent cohort of PBC patients recruited from a dedicated PBC follow-up clinic (n 114). In this validation cohort the criterion retained strongly significant predictive value for the presence of clinically significant fatigue ( 2, 30.9; P.0001). When applied to the whole study population (n 238) this predictive model generates the distributions of fatigue and its etiology as depicted in Figure 6. It should be noted that of the patients with moderate or severe fatigue, 27% remain with no identified associated etiologic process. Discussion Fatigue is an important clinical problem that adversely affects the quality of life of a significant proportion of PBC patients. Recent studies have highlighted potential associated mechanisms for fatigue in PBC (such as sleep disturbance 7 and autonomic dysfunction ). Further studies have, in contrast, highlighted important clinical processes that appear not to contribute significantly to fatigue development (eg, progressive liver disease per se and depression 3 5 ). Study of the clinical associations of fatigue in PBC, however, has been largely piecemeal to date and, in particular, there has been no attempt to integrate findings regarding the positive associations of sleep and autonomic dysfunction. This has limited the practical relevance of such research findings to clinicians who are attempting to manage patients experiencing this difficult symptom. In this study, therefore, we set out to explore the interrelationship between the symptoms of sleep abnormality and of autonomic dysfunction, and the extent to which they individually or in combination associate with fatigue in PBC. We identified the predictive criterion of ESS 10 and/or OGS 4 as being of particular value for the identification of the biological phenotypes of fatigue in PBC. In an extended clinical cohort of 238 patients moderate or severe fatigue was found to be present in 63% (thereby highlighting the impact of this symptom), 78% of whom met our proposed predictive criterion. In the derivation phase of this study we confirmed the strong association between the symptoms of daytime somnolence as

4 February 2008 FATIGUE MECHANISMS IN PBC 231 Figure 2. (A and B) Numbers of PBC patients within the derivation cohort with no fatigue (defined as scores within the range for no or mild fatigue using previously described criteria) and with fatigue (scores within the previously described range for moderate or severe fatigue) who achieve (POS) and do not achieve (NEG) the diagnostic criteria for (A) significant symptoms suggestive of daytime somnolence (ESS 10) and (B) significant symptoms suggestive of vasomotor autonomic dysfunction (OGS 4). (C) Correlation between ESS and OGS scores in individual patients. Dashed lines denote ESS 10 and OGS 4. (D) Numbers of fatigued and nonfatigued PBC patients (defined as stated earlier) within the development cohort who achieve (POS) and do not achieve (NEG) the composite predictive criterion. (A) 2, 23.8; P.0001; positive predictive value, 0.92 ( ); negative predictive value, 0.55 ( ); sensitivity, 0.47 ( ); specificity, 0.94 ( ). (B) 2, 18.2; P.0001; positive predictive value, 0.83 ( ); negative predictive value, 0.55 ( ); sensitivity, 0.53 ( ); specificity, 0.84 ( ). (D) 2, 32.0; P.0001; positive predictive value, 0.84 ( ); negative predictive value, 0.66 ( ); sensitivity, 0.71 ( ); specificity, 0.80 ( ). detected by ESS and of autonomic dysfunction as assessed by OGS and fatigue in PBC, and identified the optimal criterion of ESS 10 and/or OGS 4. Strikingly, this criterion was met by more than 95% of patients in the derivation cohort with severe fatigue and was completely absent in the patients with no fatigue. In the intermediate groups the criterion was met in approximately 20% of patients with mild fatigue and in 60% of patients with moderate fatigue. The validity of this criterion then was confirmed in a second (and, crucially, independent) clinical cohort. This cohort, derived from a specialist clinic, is likely to be very representative of the clinical problems encountered in practice. These observations led to several important conclusions. The first is that in the most profoundly affected group of patients in whom severe fatigue, typically associated with significant impairment of quality of life, is present, autonomic dysfunction and/or sleep disturbance is present almost universally. Although this study was not designed to address the direction of the relationship between autonomic dysfunction and sleep disturbance and fatigue in PBC, this finding alone would suggest that intervention to correct autonomic dysfunction and/or sleep disturbance could be of some clinical benefit to PBC patients experiencing severe fatigue, and is certainly worthy of future study. This potential is supported by recent pilot data suggesting benefit in terms of the fatigue severity after treatment with the anti-daytime somnolence drug modafinil. 8,13 The second conclusion is that a significant group of PBC patients (all, in our experience, of those patients experiencing no fatigue and the vast majority of those with low fatigue) do not show either significant daytime somnolence or autonomic dysfunction. This finding would suggest that these features are not automatic associations of the disease process but, instead, represent an additional process clearly associated with PBC but ultimately independent of it. This raises potentially interesting questions regarding the additional risk factors that must, by inference, be present in patients who go on to experience fatigue. The third important conclusion is that there is a significant minority of PBC patients with, in particular, fatigue of moderate severity who do not show either sleep disturbance or autonomic dysfunction. This finding has 2 implications. The first is that it is possible to experience significant

5 232 NEWTON ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 2 Figure 3. Percentage of PBC patients in the development cohort with PBC-40 Fatigue domain scores suggestive of no, low, moderate, and high fatigue who were identified using the fatigue etiology predictive criterion. 2, 49.6; P fatigue in PBC without showing either autonomic disturbance or sleep disturbance, a finding that would argue strongly against sleep disturbance and autonomic dysfunction as being consequences of physical activity and/or deconditioning associated with fatigue (one of the alternative potential explanations for the associations noted in this study). The second implication is that there must be at least one additional process unrelated to either autonomic dysfunction or sleep disturbance that contributes to fatigue pathogenesis in PBC. The population distributions would suggest that this additional process (or processes) is responsible for approximately 20% of moderate or severe fatigue in PBC patients. Work currently is underway to identify etiologic factors for fatigue in this currently unexplained group. Patients achieving the predictive criterion for fatigue etiology in this study did so either through the ESS component, the OGS component, or through both components. Distributions for these 3 phenotypes were approximately even across the whole cohort for both moderate and severe fatigue patients. Strikingly, however, when the severity of fatigue experienced by patients was compared in these 3 groups no significant difference was seen. In particular, those patients meeting both the ESS and OGS criteria did not have more severe fatigue than those patients meeting the diagnostic criterion for either OGS or ESS alone (all 3 groups showed Figure 4. PBC-40 Fatigue domain scores in patients in the development cohort who were positive and negative for the fatigue etiology diagnostic criterion. Median values (solid lines) are denoted for both groups. The dashed line denotes the cut-off value between mild and moderate fatigue. P Figure 5. PBC-40 Fatigue domain scores in PBC-40 patients meeting the diagnostic criterion for ESS abnormality alone (ESS, 10), OGS alone (OGS, 4), both ESS and OGS, and for neither. Median values are denoted for all groups. The solid horizontal line denotes the mild/ moderate fatigue boundary and the dashed lines denote the none/mild and moderate/severe boundaries. ***P.0001 versus neither. substantially greater fatigue severity than the group of patients who met neither of the criteria). This observation both raises interesting questions regarding fatigue pathogenesis, and has important implications for potential therapeutic application of these data. The implication for etiology is that fatigue appears to be subject to some form of threshold effect in PBC and that significant autonomic dysfunction or sleep disturbance in isolation is sufficient to reach this threshold. The presence of the second factor does not then appear to add to the fatigue, even if the threshold for effect also os reached. Further mechanistic work in this area clearly is needed. The important implication for therapeutics is, of course, that it would seem likely that patients meeting both of the criteria in whom one of the etiologic factors (autonomic dysfunction or sleep disturbance) is treated might be predicted to not improve because the presence of the nontreated component would then predominate which itself is associated with ongoing fatigue. It is important, therefore, in the future design of studies that address the treatment of autonomic dysfunction or sleep disturbance and its impact on fatigue in PBC to quantify the other factors that may contribute to fatigue. Protocols must, therefore, restrict themselves to patients in whom a single fatigue etiologic factor is apparently at play. Taken together our findings suggest that a significant proportion of fatigue in PBC is associated with the presence of autonomic dysfunction, sleep disturbance, or both. Potential interventions for each of these processes have been described (fluid management, salt intake management, fludrocortisone and midodrine for autonomic dysfunction, and avoidance of sleep-reducing factors such as caffeine, ventilatory support, and modafinil in the case of daytime somnolence), which are clearly worth studying in the context of PBC patients with fatigue (stratified as described earlier). Only time will tell whether the specific interventions aimed at addressing autonomic dysfunction and sleep disturbance truly do benefit fatigue, however, we believe that there is now sufficient evidence for these approaches to be tested in clinical practice. We would argue that such approaches are needed urgently considering the recently recognized associations between fatigue in PBC and excess mortality. 18,19

6 February 2008 FATIGUE MECHANISMS IN PBC 233 Figure 6. Distribution of fatigue severity and its causes/associated features in the whole study cohort. References 1. James OFW, Bhopal R, Howel D, et al. Primary biliary cirrhosis once rare, now common in the UK? Hepatology 1999;30: Kaplan MM, Gershwin ME. Primary biliary cirrhosis. N Engl J Med 2005;353: Cauch-Dudek K, Abbey S, Stewart DE, et al. Fatigue in primary biliary cirrhosis. Gut 1998;43: Poupon RE, Chretien Y, Chazouilleres O, et al. Quality of life in patients with primary biliary cirrhosis. Hepatology 2004;40: Huet PM, Deslauriers J, Tran A, et al. Impact of fatigue on the quality of life in patients with primary biliary cirrhosis. Am J Gastroenterol 2000;95: Newton JL, Bhala N, Burt J, et al. Characterisation of the associations and impact of symptoms in primary biliary cirrhosis using a disease specific quality of life measure. J Hepatol 2006;44: Newton JL, Gibson JG, Tomlinson M, et al. Fatigue in primary biliary cirrhosis is associated with excessive daytime somnolence. Hepatology 2006;44: Jones DEJ, Newton JL. An open label study of Modafinil for the treatment of excessive daytime sleepiness and fatigue in primary biliary cirrhosis. Aliment Pharmacol Ther 2007;25: Keresztes K, Istenes I, Folhoffer A, et al. Autonomic and sensory nerve dysfunction in primary biliary cirrhosis. World J Gastroenterol 2004;10: Newton JL, Hudson M, Tachtatzis P, et al. Population prevalence and symptom associations of autonomic dysfunction in primary biliary cirrhosis. Hepatology 2007;45: Newton JL, Allen J, Kerr S, et al. Reduced heart rate variability and baroreflex sensitivity in primary biliary cirrhosis. Liver Int 2005;26: Newton JL, Davidson A, Kerr S, et al. Autonomic dysfunction in primary biliary cirrhosis correlates with fatigue severity. Eur J Gastroenterol Hepatol 2007;19: Kaplan MM, Bonis PAL. Modafinil for the treatment of fatigue in primary biliary cirrhosis. Ann Intern Med 2005;143: Goldblatt J, Taylor PJS, Lipman T, et al. The true impact of fatigue in primary biliary cirrhosis: a population study. Gastroenterology 2002;122: Jacoby A, Rannard A, Buck D, et al. Development, validation and evaluation of the PBC-40, a disease specific health related quality of life measure for primary biliary cirrhosis. Gut 2005;54: Schrezenmaier C, Gehrking JA, Hines SM, et al. Evaluation of orthostatic hypotension: relationship of a new self-report instrument to laboratory-based measures. Mayo Clin Proc 2005;80: Johns M. Sleepiness in different situations measured by the Epworth Sleepiness Scale. Sleep 1994;17: Jones DEJ, Bhala N, Burt JA, et al. Four year follow up of fatigue in a geographically defined primary biliary cirrhosis patient cohort. Gut 2006;55: Jones DEJ, Newton JL. Fatigue-associated mortality in primary biliary cirrhosis is principally a problem of the over 65 s. Gut 2007;56:1166. Address requests for reprints to: Professor David E. J. Jones, Liver Research Group, Institute of Cellular Medicine, 4th Floor William Leech Building, The Medical School, Framlington Place, Newcastle-upon- Tyne, NE2 4HH, United Kingdom. D.E.J.Jones@ncl.ac.uk; fax: (44) This work was funded by the Medical Research Council.