Portal de Periódicos da CAPES

Noninvasive monitoring for methotrexate hepatotoxicity

2005; Oxford University Press; Volume: 152; Issue: 3 Linguagem: Inglês

10.1111/j.1365-2133.2005.06605.x

1365-2133

A. M. MacDonald, A. David Burden,

Immunotoxicology and immune responses

The advent of targeted biological treatments for psoriasis has challenged the safety and efficacy of established systemic treatments. Methotrexate has been used for psoriasis since the late 1950s and is still the most widely prescribed drug for severe psoriasis worldwide. Its effectiveness in treating psoriasis was first detected by chance in a patient being treated for arthritis whose coincident psoriasis was observed to improve.1 This started a trend of serendipity in drug development for psoriasis, being subsequently followed by ciclosporin, vitamin D and infliximab. As methotrexate has been in use for almost 50 years it predates the development of randomized controlled trials, and high-quality data concerning its efficacy are sparse. Consequently there is considerable variability in the recommended weekly dose, the route of administration (oral, intramuscular, intravenous and subcutaneous) and the need for, and dosing of, concomitant folic acid. With improvements in our understanding of the immunopathogenesis of psoriasis, it has become apparent that the mechanism of action in psoriasis (originally attributed to the inhibition of the enzyme dihydrofolate reductase, disrupting DNA synthesis and interrupting epidermal cell division) is probably related to effects on T cells, rather than keratinocytes.2, 3 Use of methotrexate is restricted by the risk of organ toxicity. Major concerns are myelosuppression, hepatotoxicity and pulmonary fibrosis, of which myelosuppression is the most important. Of the 164 fatalities associated with methotrexate therapy reported to the U.K. Committee on the Safety of Medicines (CSM) between 1969 and 2004, 67 were associated with myelosuppression. In contrast, 30 deaths occurred in association with pulmonary complications and only eight as a result of liver complications (CSM, personal communication, 2004). It is possible that there is under-reporting of cases to the CSM, as a survey in 1992 of 278 U.K. dermatologists treating approximately 10 000 patients with methotrexate over a mean period of 12 years reported 23 deaths as a result of liver failure and five as a result of liver biopsy.4 Prescribing error is a particular problem and this may increase in the U.K. with a shift in monitoring of methotrexate to primary care. The National Patient Safety Agency recorded 137 patient safety incidents in the U.K. over the last 10 years, including 25 deaths and 26 cases of serious harm to the patient as a consequence of methotrexate administration (http://www.npsa.nhs.uk, accessed 21 December 2004). Two-thirds of incidents were prescribing errors, 19% were due to a lack of, or poor monitoring, and 7% were due to misidentification of tablets by professionals or patients. Most incidents involved primary care clinicians. Among dermatologists, there is particular concern about the long-term hepatotoxic potential of methotrexate. Interestingly, this concern is not shared to the same extent by other specialities.5 It seems quite possible that the risk of hepatotoxicity varies in different diseases.6 For instance, patients with rheumatoid arthritis are more frequently treated with oral corticosteroids than are patients with psoriasis, and it is conceivable that this reduces hepatic inflammation. Psoriatic patients have a greater tendency to drink alcohol to excess, which may contribute. Among patients with psoriasis receiving methotrexate, rates of hepatic fibrosis as high as 25–50% have been published6-8 athough recent estimates are lower, at 0–4%.9, 10 The cause of methotrexate-induced fibrosis is unknown. Methotrexate may also be an independent risk factor for nonalcoholic steatohepatitis (NASH),11 the most common cause of chronic liver disease in the developed world. Folate supplementation reduces the frequency of raised liver enzymes without altering efficacy,12 suggesting that this aspect of methotrexate toxicity may be divorced from its therapeutic efficacy. Understanding of liver fibrosis and cirrhosis has advanced significantly over recent years. Hepatic stellate cells (HSCs) are now known to be the final common pathway in the production of collagens. In hepatic fibrosis, HSCs are upregulated to become myofibroblast-like and produce collagen. Gliotoxin, which stimulates apoptosis of HSCs, promotes reversal of liver fibrosis in animal models.13 Research into the mechanisms of activation of HSCs by chemokines and cytokines may allow for therapeutic intervention in the future.14 Understanding of the role of metalloproteinases, which degrade collagen, may also provide a therapeutic target to reverse fibrosis.14 Ideally, individuals at particular risk of hepatotoxicity would be identified prior to treatment. Early pharmacogenetic studies have demonstrated an A1298C polymorphism in the methylenetetrahydrofolate reductase gene which may protect against methotrexate-related adverse effects,15 while a C667T polymorphism in the same gene correlates with methotrexate hepatotoxicity.16 These studies assessed tolerability of methotrexate and derangement of liver enzymes rather than progression to fibrosis, but hopefully this aspect will also be addressed. Environmental factors predisposing to methotrexate hepatotoxicity include exposure to other hepatotoxins such as alcohol, and also the various factors associated with NASH, such as obesity and diabetes.17 Many issues need to be considered in deciding a policy for screening for methotrexate hepatotoxicity, but it is important that the procedures used are not only accurate, but are also proportionate to the frequency and severity of this complication. It is widely accepted that elevations in liver enzymes are a poor guide to the development of liver fibrosis. However, although liver enzymes rise in the days following methotrexate dosing, the significance of this for the subsequent development of fibrosis is not known. It is recommended that liver enzymes be measured at least 7 days after the weekly dose of methotrexate so that this transient rise is not detected.17 Unfortunately, the timing of liver enzyme measurement in relation to dosing is often not recorded either in clinical practice or in scientific papers. In one of the few randomized controlled trials of methotrexate in psoriasis, 12 of 43 patients receiving methotrexate were withdrawn due to transient increases in liver enzymes, although the timing of blood sampling was not mentioned.18 Various modalities of imaging the liver have been evaluated. Ultrasound19, 20 and magnetic resonance imaging21 are ineffective for detecting fibrosis, and although dynamic scintigraphy may detect the presence of liver damage, it does not predict the severity.22 Therefore, liver biopsy is usually recommended as the gold standard for detecting methotrexate-induced hepatic fibrosis. Current American Academy of Dermatology (AAD) guidelines suggest an early liver biopsy in those with risk factors, and biopsy for all patients at a cumulative dose of 1·5 g and thereafter at 1–1·5 g intervals.17 However, needle biopsy samples only 1/50 000 of the liver23 and is conceptually similar to a blind biopsy of the skin in widespread skin disease: the sampling error is about 30%.24 The information gained is static and morphological rather than dynamic and functional. Routine liver biopsy results influence management in a minority of cases,9, 25 are unpopular with patients, and are associated with significant morbidity and occasional mortality. For all these reasons, noninvasive methods of detecting and monitoring liver fibrosis are highly desirable and the requirement for liver biopsy is being questioned.26 The aminoterminal peptide of type III procollagen (PIIINP) is produced during the synthesis of type III collagen. Serum PIIINP correlates directly with the amount of ongoing hepatic fibrogenic activity but not with previous liver damage. Over the past 20 years, serial measurements of serum PIIINP have been found to correlate with the presence and severity of liver fibrosis in several clinical settings, including methotrexate treatment of psoriasis.27-29 In this issue of the BJD, two studies of the effect of PIIINP measurement in reducing the need for liver biopsy are reported. Chalmers et al.30 describe an audit carried out over 24 months in four U.K. centres in which different protocols were used for monitoring patients receiving methotrexate. One hundred and sixty-six patients had PIIINP measured every 2–3 months and a liver biopsy was offered only to those with a high pretreatment PIIINP, elevated PIIINP above the normal range in three samples over a 12-month period or elevation of PIIINP above 8 µg L−1 in two consecutive samples. Eighty-seven control patients were managed according to the AAD guidelines described above. A sevenfold reduction in the need for liver biopsy was observed in the PIIINP group compared with the AAD guideline group. Abnormalities in the liver biopsy of sufficient severity to influence management were identified in 20% of the intervention group and 6% of the control group. Ninety-five per cent of patients stated a preference for serial PIIINP measurement over liver biopsy. In a second paper, Maurice et al.31 report the ability of PIIINP levels to identify those with and without hepatic fibrosis in a cohort of 34 patients. Thirteen per cent of patients had fibrosis, this group having received a higher cumulative dose of methotrexate. In 46 liver biopsies, histology was compared with the median PIIINP level where this had been checked within 12 months of the biopsy. A consistent increase in PIIINP level was seen across each grade of histology. No biopsy showed fibrosis if all associated PIIINP assays had been normal, but 50% of liver biopsies without fibrosis had at least one abnormal PIIINP assay. In patients who had undergone sequential liver biopsies, histology was compared with serial PIIINP levels. Concordance was found in all four biopsy pairs showing fibrosis, with elevated levels of PIIINP in over 50% of intervening assays. However, 63% of stable biopsy pairs had at least one abnormal PIIINP assay. Avoiding liver biopsies in patients with consistently normal PIIINP levels would have resulted in a 45% reduction in biopsies. These studies, combined with the previously published literature, suggest that serial PIIINP assays are sensitive but not specific in detecting hepatic fibrosis. PIIINP is not organ specific, and may be raised in children and in various pathological states including arthritis, scleroderma and hyperthyroidism. There is a need to standardize the level of PIIINP that necessitates a liver biopsy. Nevertheless, it would seem reasonable to suggest that in the absence of elevated serum PIIINP levels the risk of significant hepatic fibrosis is very low and a liver biopsy need not be routine. Dermatologists in Europe are starting to adopt this practice. In a survey reported in 2004, 43% of a sample of dermatologists in Ireland stated that they perform liver biopsies on the basis of PIIINP level and 30% according to cumulative methotrexate dose.5 There remains a need for more specific noninvasive markers in those patients with elevated serum PIIINP levels, many of whom do not have hepatic fibrosis. A variety of methods has been investigated in other diseases, particularly hepatitis C virus infection. These include indirect markers of liver fibrosis such as α2-macroglobulin, α2-globulin, γ-globulin, apolipoprotein A1, γ-glutamyltransferase and total bilirubin.23 Results can be combined to produce calculated indices (for example the 'FibroTest') which have been applied to the detection of fibrosis with varying degrees of success.32, 33 There is also interest in direct markers of fibrosis in addition to PIIINP, including procollagen I, type IV collagen, laminin, hyaluronic acid (HA), tissue metalloproteinases and their inhibitors. Sensitivity and specificity vary depending on the cause of fibrosis.23 For instance, HA has been shown to have greater diagnostic accuracy than PIIINP in viral and alcoholic liver disease34-36 and in primary biliary cirrhosis.37 As each direct marker measures a different aspect of the fibrogenic/fibrolytic process, panels of markers have recently been investigated in prospective clinical trials.38 This approach remains untested in detecting methotrexate hepatotoxicity but might be expected to improve the positive predictive value of the test compared with PIIINP alone. It seems likely that methotrexate will retain a key role in the management of psoriasis despite the therapeutic advances offered by targeted biological therapy.10 For many patients methotrexate remains an attractive therapeutic option, and it would be more appealing if the minority of patients who develop hepatic fibrosis could be predicted by pharmacogenetics, detected reliably and noninvasively at an early stage, and treated effectively. These aspirations may soon be a reality, and the development of PIIINP monitoring is a step in this direction.