Diclofenac-Associated Hepatotoxicity: Analysis of 180 Cases
`Reported to the Food and Drug Administration
`as Adverse Reactions
`
`
`
`ALPHA T. BANKS,’ H ~ A N J . ZIMMERMAN,’’~ &VAL G. ISHAK,~ AND JOHN G. HARTER’
`
`Diclofenac is a nonsteroidal anti-inflammatory drug
`approved in the United States in 1988 for the treatment
`of patients with osteoarthritis, rheumatoid arthritis, or
`ankylosing spondylitis. To characterize the clinical, bio-
`chemical, and histological features and possible mecha-
`nisms of hepatic injury associated with its use, a retro-
`spective analysis was undertaken of 180 patients whose
`cases were reported to the Food and Drug Administra-
`tion from November 1988 through June 1991, as having
`had possible adverse reactions to diclofenac. Of the re-
`ported 180 cases, 79% were female, 71% were 60 years
`of age or older, and 77% had osteoarthritis. Sixty-seven
`percent of the cases were detected by symptoms and the
`remainder by abnormal laboratory tests. Seventy-five
`percent of the symptomatic patients (90 of 120) were
`jaundiced. Seven of the 90 icteric patients died. The bio-
`chemical pattern of injury was hepatocellular or mixed
`hepatocellular in 66% of cases. Only 8% had a pattern of
`cholestatic injury. The remainder, with modestly in-
`creased values of both transaminases and alkaline phos-
`phatase, were considered “indeterminate,” i.e., either
`mild hepatocellular or anicteric “cholestatic” injury.
`Sections of liver from 21 cases were available for study.
`Hepatic injury was apparent by 1 month after starting
`the drug in 24%. by 3 months in &3%, and by 6 months
`in 85% of cases. The latent period in 12% was 6 to 12
`months, whereas in 370 it was greater than 12 months.
`A combination of rash, fever, and eosinophilia, all hall-
`marks of immunological idiosyncrasy (hypersensitivity),
`was not reported in any case; additionally, the long la-
`tent period in most of the patients led to the inference
`that the mechanism is probably metabolic idiosyncrasy.
`The data suggest that diclofenac-related liver injury is
`particularly likely to involve osteoarthritic females,
`presenting with jaundice 1 to 6 months after starting
`diclofenac, with injury that is predominantly hepatocel-
`
`Abbreviations: NSAID, nonsteroidal anti-inflammatory drug; KA, rheuma-
`toid arthritis; OA, osteoarthritis; FDA, Food and Drug Administration; AST,
`aspartate transaminase; AI.T, alanine transaminase; AI.P, alkaline phospha-
`tase; ULN. upper limit of normal.
`From the Food and Drug Administration. George Washin@on University,
`and the Armed Forces Institute of Pathology, Washington, DC.
`Received July 8, 1994; accepted May 2, 1995.
`Address reprint requests to: Hyman d. Zimmerman, MD, Armed Forces
`Institute of Pathology, Department of Hepatic and Gastrointefitinal Pathology,
`14th and Alaska Ave, NW, WashinbTon, DC 20306-6000.
`Copyright Q 1995 by the American Association for the Study of 1.iver
`Diseases.
`0270-9 i39/95/2203-ooi~$~.0010
`
`lular and presumably caused by metabolic idiosyncrasy.
`(HEPATOLOGY 1995;22:820-827.)
`
`Diclofenac is a nonsteroidal anti-inflammatory drug
`(NSAID) approved in the United States in 1988 for
`treatment of rheumatoid arthritis (RA), osteoarthritis
`(OA), and ankylosing spondylitis’ and in 1993 for the
`management of pain. The drug has been available in
`Europe since 1974. As is true of other NSAIDs, diclo-
`fenac has been associated with gastric hemorrhage, re-
`nal dysfunction, serious hepatic injury, and blood
`dyscra~ia.‘-~~ Despite the attributedzs uniformity of
`NSAIDs with regard to adverse hepatic effects, there
`appear to be differences among them in the hepatic
`injury that they may provoke, including its character,
`presumed mechanism, estimated incidence, and clini-
`
`cal i m p ~ r t a n c e . ~ ~ - ~ ’ Accordingly, the characterization of
`the hepatic injury caused by individual NSAIDs seems
`warranted. A previous study, similar to the present
`one, focused on sulindac-associated hepatic injury.3’
`Approximately 60 cases of diclofenac-associated he-
`patic injury have been reported in the medical litera-
`ture; most have shown acute hepatocellular injury.2-21
`Several have shown a pattern of chronic hepati-
`tis. 14.16,22,23
`The individual reports have varied in the
`completeness of the data and the light that they shed
`on factors affecting susceptibility to injury and its char-
`acter. Accordingly, the effort was undertaken to define
`more fully the character of the injury and to search for
`factors affecting susceptibility among a large number
`of suspected cases of diclofenac-related liver injury re-
`ported to the spontaneous reporting system of the Food
`and Drug Administration (FDA). After follow-up to es-
`tablish the validity of the data, the reports were used to
`characterize the biochemical pattern of injury, clinical
`features, and histological changes, as well as the mech-
`anisms of injury.
`
`MATERIALS AND METHODS
`There were 434 reports of diclofenac-associated hepatic in-
`jury submitted to the voluntary reporting system of the FDA,
`directly or through the manufacturer, between November
`1988 and June 1991. These cases were identified for review
`by using Coding Symbols for Thesaurus of Adverse Reaction
`Terms suggestive of hepatic in jury.:"'^:" As was done in the
`study of sulindac-associated injury,“’ the validity of the data
`820
`
`

`
`HEPATOLOGY Vol. 22, No. 3, 1995
`
`BANKS ET AL 821
`
`Aminotransferase
`Fold-elevated
`
`> 8x
`
`c 8x
`
`Hepatocellular Injury
`
`Mixed Injury
`22 patients (12%)
`Jaundice
`19 patients
`
`Jaundice
`60 patients
`Cholestatic Injury
`Indeterminate Injury
`8 patients (4%)
`46 patients (26%)
`- _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
`Cholestatic Jaundice
`Cholestatic
`(Hepatocanalicular
`Jaundice
`(Canalicular Type)
`Type)
`4 patients
`7 patients (4%)
`c 3x
`
`Alkaline Phosphatase
`Fold-elevated
`
`FIG. 1. Graphic presentation of criteria for types of hepatic in-
`jury. See text for description. Note that serum transaminase levels
`less than eight times the upper limit of normal with ALP less than
`three times the normal define canalicular cholestasis if there is jaun-
`dice or bilirubin values greater than 2.4 mg% or indeterminate injury
`if neither jaundice nor bilirubin value greater than 2.4 is present.
`
`later than 5 weeks after starting the drug was presumed to
`reflect metabolic idiosyncrasy.
`In attempting to identify the effect of age, gender, and
`indication for diclofenac therapy on susceptibility, the propor-
`tion of each in the reports was compared with the number of
`prescriptions for the respective category (information ob-
`tained from the IMS American, Inc National Prescription
`Audit and the IMS National Drug Therapeutic Index for the
`period from January 1988 to June 1991 [Table 21). Propor-
`tions are expressed as percentages. Statistical significance
`was evaluated by the x2 test.
`RESULTS
`A total of 180 cases of hepatic injury were analyzed
`(Table 1). In 89% of the cases the drug had been taken
`for treatment of rheumatoid arthritis or osteoarthritis
`(Table 11, yet only 34% of the use of the drug was for
`these conditions, and 66% was for conditions other than
`
`~
`
`~~
`
`TABLE 1. Cases of Diclofenac Hepatic Injury (n = 180)
`Gender
`Females
`Males
`Age (yr)
`< 40
`40-59
`>60
`Missing
`Indications for use of diclofenac:
`Osteoarthritis
`Rheumatoid arthritis
`Other
`Ankylosing spondylitis
`Psoriatic arthritis
`Miscellaneous conditions
`Unknown
`
`142
`38
`
`8 (5%)
`43 (24%)
`123 (68%)
`6 (3%)
`
`139
`22
`19
`1
`1
`15
`2
`
`was established by written inquiry and telephone calls to
`the reporting physician by one of the authors (ATB). The
`manufacturer (Ciba-Geigy, Basel, Switzerland) also supplied
`available clinical information on cases on which it had data.
`Each case was reviewed by the authors to evaluate the valid-
`ity of the relationship to diclofenac, to categorize the form of
`injury, and to deduce the apparent mechanism from the clini-
`cal features.
`After elimination of 254 cases because of duplicate re-
`porting, foreign sources, other possible causes for liver
`enzyme abnormalities, and inability to obtain adequate in-
`formation, 180 cases remained for analysis (Table 1). Bio-
`chemical data to characterize liver injury (bilirubin, aspar-
`tate transaminase [ASTI, alanine transaminase [ALT], and
`alkaline phosphatase [ALP]) were available or solicited. They
`included the normal values for the respective laboratory,
`baseline levels, and values during the acute event and, when
`available, previous and subsequent values. The pattern of
`injury was classified as hepatocellular, cholestatic, mixed, or
`indeterminate (Fig. 1). Hepatocellular injury was defined by
`a peak transaminase increase of at least eight times the up-
`per limit of normal (ULN) with an ALP value less than three
`times the ULN, and cholestatic injury was defined (hepato-
`canalicular type3') by an ALP value at least three times the
`ULN and transaminase values of less than eight times the
`ULN. "Mixed" injury was defined by transaminase increases
`greater than eight times the ULN accompanied by an ALP
`value of more than three times the ULN. The combination
`of transaminase increases less than eight times the ULN
`with ALP values less than three times the ULN was classified
`as cholestatic (canalicular type3') if jaundice was present or
`serum bilirubin levels were above 2.4 mg/dL and as indeter-
`minate in the absence of these markers of hyperbilirubi-
`nemia. Serological tests that were available for review in
`most cases included hepatitis B surface antigen, antibody to
`hepatitis B core antigen (immunoglobulin M), and antibody
`to hepatitis A virus (immunoglobulin M). Results of a test
`for hepatitis C (Anti HCV) were available in only 19% of the
`patients.
`Presumption of mechanism of injury was based on clinical
`criteria." Presence of fever, with or without rash, andor eo-
`sinophilia, if accompanied by abrupt onset during the first 4
`to 5 weeks of treatment, was presumed to reflect immunologi-
`cal idiosyncrasy (hypersensitivity). Lack of clinical hallmarks
`of hypersensitivity andor gradual development of the illness
`
`TABLE 2. January 1988 to June 1991 Drug Mentions
`for Diclofenac (in millions)
`OA + RA
`RA
`AllOthers
`4.80
`1.52
`3.27
`2.07
`0.77
`1.29
`6.86
`
`Total
`
`12.94
`6.15
`6.80
`7.36
`4.22
`3.14
`20.30
`
`8.15
`4.62
`3.53
`5.29
`3.45
`1.85
`13.44
`
`OA
`
`3.52
`0.86
`2.66
`1.48
`0.45
`1.03
`5.01
`
`1.27
`0.66
`0.61
`0.58
`0.32
`0.26
`1.86
`
`Female
`0-59
`60+
`Male
`0-59
`60+
`Total
`
`NOTE. Source: National Disease and Therapeutic Index IMS
`American, Inc.
`Abbreviations: OA, osteoarthritis; RA, rheumatoid arthritis.
`
`Page 2
`
`

`
`822 BANKS ET AL
`
`TAHLE 3. Diclofenac Hepatic Injury in Patients With RA or
`OA: Comparison of Expected With Observed Distribution of
`Cases According to Gender, Age, and Diagnosis
`Use (in
`millions)
`
`Relative Risk
`
`Injured
`
`~~
`
`~
`
`4.80 (70%)
`2.07 (30%)
`6.87
`2.02
`
`132 (82%)
`29 (185%)
`161
`4.55
`
`4.56 (67%)
`2.29 (33%)
`6.85
`1.99
`
`6
`112 (724)
`43 (28% )
`155
`2.6
`
`2.0 (I’ < ,001)
`
`1.3 (P = ,131
`
`Gender
`Femalc
`Male
`Total
`Ratio (F-MJ
`At9
`(missing)*
`2 60
`<60
`Total
`Ratio (z60:<60)
`Diagnosis
`OA
`RA
`Total
`Ratio (OA-RAJ
`
`5.01 (72%)
`1.86 (28% J
`6.87
`2.69
`
`139 (86C7r)t
`22 (14% )+
`161
`6.32
`
`2.4 ( P = ,001)
`Abbreviations: RA, rheumatoid arthritis; OA, osteoarthritis.
`The 6 missing values did not enter the corresponding calcula-
`tions of ratios or percentages.
`+Figure in parentheses refers to proportion of all patients with
`arthritis.
`
`arthritis (Table 2). Although information regarding du-
`ration of use was not available, the variety of condition
`other than arthritis may be presumed to have been
`treated for short periods in contrast to the usual long-
`term treatment of patients with arthritis. Distribution
`by gender, diagnosis, and age for both cases and users
`is shown in Table 3. The ratio of females to males
`among cases with hepatic injury was 4.55 compared
`with the gender ratio of 2.32 among users, yielding a
`relative risk of 2.0 for females ( P < .001, Table 3). Age
`appeared to have no identifiable effect. When the ratio
`of individuals aged 60 years or older to those under 60
`years old was compared with the ratio among users,
`the relative risk was 1.3 (P = .13). The 6.32 ratio of
`osteoarthritis to rheumatoid arthritis among cases,
`compared with the 2.69 ratio among users of the drug,
`yielded a relative risk of 2.4 (P < .001; Table 3). The
`small number of cases with other diagnoses and the
`heterogeneity of the group precluded useful analysis.
`Two thirds of the cases (120) presented with signs or
`symptoms. The remainder were identified by increased
`levels of AST and ALT noted by monitoring (18%) or
`as incidental tests (15%) (Fig. 2). Jaundice was present
`in 90 of the 180 cases (50%) (Fig. 3). It was the only
`complaint in 14 of the 90 icteric cases (17%) and was
`accompanied by anorexia, nausea, and/or vomiting in
`58 of the 83 non-fatal icteric patients (70%). These
`symptoms also occurred in 21 of the 29 anicteric symp-
`tomatic patients (72%). Abdominal discomfort was re-
`corded in 19 (16%) of the 120 symptomatic patients,
`and pruritus was noted in 15 of them (12%). Seven of
`the 90 icteric patients died, yielding a case fatality rate
`
`HEPATOI.OGY September 1995
`D i cl of en a
`2 Hepatic Injury
`Methoc
`of Detection
`
`Symptoms (67%) n
`
`Monitoring (18%)
`
`FIG. 2. Circumstances that brought cases to attention
`
`for icteric cases of 8%. There were no fatalities among
`the anicteric cases.
`The duration of therapy before the detection of he-
`patic injury was under 6 months in 85% of all cases
`(<1 month [24%1; 1 to 3 months [39%l; 3 to 6 months
`l22%I). In 12% of all cases, the duration of treatment
`before the detection of injury ranged from 6 to 12
`months. In 3% of all cases, more than 1 year had
`elapsed before evidence of hepatic injury was detected
`(Fig. 4).
`Specimens of hepatic tissue were available from 21
`cases. All but 6 disclosed hepatocellular injury (Fig. 5A
`through 5C). One case showed a mixed hepatocellular-
`cholestatic injury, and 2 cases showed mainly cholesta-
`sis with portal area inflammation and cholangitis (Ta-
`ble 4). Six of the cases of hepatocellular injury revealed
`changes of chronic hepatitis (Fig. 5D), 4 with consider-
`
`A.N.V
`14 __
`-_
`7
`0
`
`No Jaundb
`
`Asym
`
`90
`
`A.N.V
`
`A.N.V.F
`
`Other
`
`Jaundice Only
`
`83
`
`FIG. 3. Clinical manifestation of 120 symptomatic patients, 90
`icteric patients, and 30 of the 90 anicteric patients. Abbreviations:
`A, anorexia; N, nausea; V, vomiting; F, fever; Other, abdominal pain,
`pruritus, and/or malaise; Asym, asymptomatic.
`
`Page 3
`
`

`
`Trl
`
`97
`
`100
`
`~
`
`05 ___
`
`79
`
`~
`
`73
`
`63
`
`HEPATOLOGY Vol. 22, No. 3, 1995
`
`Percentage
`120
`
`mi
`
`1 00
`
`80
`
`I
`
`40
`
`24
`
`20
`
`0
`
`1
`
`2
`
`3
`
`4
`5
`Months
`
`6
`
`__
`12
`
`~ > 12
`
`FIG. 4. Duration of intake of diclofenac before appearance of he-
`patic injury. Cumulative figures for onset of injury by end of each
`period.
`
`able lobular injury and 1 with cirrhosis. Nine showed
`acute hepatocellular injury with spotty necrosis in 2
`cases, zone 3 necrosis in 6 cases, and granulomatous
`inflammation in 1 case. In the remaining 3 cases, the
`injury was trivial or nonspecific.
`Of the 180 cases, 119 (66%) had increased transam-
`inase values to more than 8 times the ULN (Fig. 1).
`Among the 90 icteric cases, 88% (79 cases) had trans-
`aminase levels in that range. In three fourths of cases
`(60 patients) the pattern was that of hepatocellular
`injury with ALP levels less than three times the
`ULN; and in the remainder (19 patients) the injury
`was mixed with ALP levels more than three-fold in-
`creased (Fig. 1). Only 8 patients (4%) had ALP in-
`creases more than 3 times the ULN with transami-
`nase values
`less
`than eight
`times
`the ULN
`(cholestatic [hepatocanalicular type] pattern), and
`29% of all cases had transaminase and ALP values
`of the indeterminate pattern-either mild hepatocel-
`lular, i.e., anicteric (26%) or cholestatic (canalicular
`type) injury, i.e., icteric or bilirubin level greater
`than 2.4 mg/dL (4%).
`Hallmarks of hypersensitivity were uncommon.
`None of the patients had fever, rash, and eosinophilia.
`Fever and rash were present in only 2 patients. Fever
`alone was recorded in 17 additional patients and rash
`alone in 5 others. Eosinophilia was observed in 7 other
`patients, all without fever or rash. Readministration
`of the drug led to prompt recurrence of abnormality in
`1 of 19 patients tested. The remaining 18 developed
`increased ALT levels as early as 5 days and as late as
`1 year after restarting the drug.
`There were seven fatal cases; all had been jaundiced.
`Six of the fatal cases were female, six were older than
`60 years of age, and all seven had osteoarthritis. Peak
`transaminase values were usually in excess of 1,000
`IUD,, and bilirubin levels ranged from 13 to 25 mg/dL.
`The duration of therapy for the fatal cases ranged from
`4 to 69 days (median of 24 days). An eighth patient with
`mild hepatic injury died of a complication of subclavian
`
`BANKS ET AL 823
`
`artery catheterization while under treatment for renal
`failure.
`
`DISCUSSION
`Attribution to diclofenac of the etiologic role in the
`hepatic injury of the 180 cases among the 434 cases
`reported to the FDA was based on excluding cases with
`other recognized possible etiology. Hepatitis A and B
`were excluded by serological tests in 96 of the patients
`(53%). Thirty-four patients (19%) had negative tests
`for hepatitis C. Although testing for hepatitis C is of
`limited help in diagnosis of acute hepatitis, there was
`little reason to consider that entity to be the diagnosis
`in the 180 cases, i.e., no history of recent transfusion
`or drug abuse. There were no reports of tests for human
`immunodeficiency virus.
`The hepatic injury induced by diclofenac, reflected
`by the biochemical values, is mainly hepatocellular.
`The hepatocellular character of the injury is also re-
`flected in the 8% case fatality rate for icteric cases since
`it is characteristic of drug-induced hepatocellular jaun-
`dice to have a case fatality rate in that range.32 The
`histological character of the injury in the 21 cases with
`material for study also shows the main thrust to be
`hepatocellular. These observations are consistent with
`most other reports.1-22 However, cholestatic injury also
`has been reported13 and was found in 8% of our patients
`(Fig. l), hepatocanalicular cholestasis in 4, and canalic-
`ular cholestasis in 7 patients. Chronic hepatitis has
`been reported by other^^^,^^ and was seen in 6 of the
`21 patients with histology in this study.
`The mechanism of injury appears to be idiosyncrasy
`rather than intrinsic toxicity of the drug, since the inci-
`dence is very low. In view of the rarity of hallmarks of
`hypersensitivity and the delayed development of injury
`(after more than one month of taking the drug) in most
`cases (76%), and the delayed response to rechallenge
`in all but one of the 19 patients who responded to the
`readministration, the inference of metabolic rather
`than immunologic idiosyncracy seems tenable.32 Other
`observers also have drawn attention to the rarity of
`features of hypersensitivity." Furthermore however,
`Shapiro et al' have reported signs of hypersensitivity
`in association with diclofenac injury. In these few
`cases, the mechanism may have been immunological
`idiosyncrasy (hypersensitivity) despite lack of other
`hallmarks.
`The likelihood that metabolic idiosyncrasy may be the
`mechanism is supported by the observation that diclo-
`fenac can injure hepatocytes in uitro33-36 and that inhibi-
`tion of metabolism inhibits the injury.36 Although the acyl
`glucuronide of diclofenac binds covalently to hepato-
`
`c y t e ~ , ~ ~ , ~ ~ its role in causing hepatic injury is not
`Metabolic transformation of diclofenac involves forma-
`tion of hydroxy derivatives of the benzyl nucleus.35 In
`the process, reactive toxic metabolites could be formed.
`
`Recent studies by Boelsterli and Kretz-R~mmel~~ have
`shown that diclofenac-treated hepatocytes carry anti-
`genic determinants that can be recognized by T-cell and
`by B-celVmacrophage combinations from diclofenac-
`
`Page 4
`
`

`
`PIC. 5. (A) Biopsy section ol' case showing acute hepatocellular injury. There is drop out of cells in zone 3 (arrow). Adjacent, relatively
`spared livcr cells show ballooning degeneration. (Hematoxylin-eosin; original magnification x 120.) (B) Same case showing zone 3 necrosis
`that is outlined by hypertrophied Kuplfer cells containing lipofuscin (black). (Periodic acid-Schiff after diastases digestion, original magnifica-
`tion X300.) (C) Autopsy section of case with massive necrosis. Almost all liver cells have dropped out and the residual stroma is congested.
`Note portal areas (top and bottom). (Hematoxylin-eosin; original magnification X7.5.) (D) Biopsy section of case showing chronic hepatitis.
`A markedly expanded portal arca is infiltrated with numerous inflammatory cells. The junction of the expanded portal area and adjacent
`parenchyma is ill defined (piecemeal necrosis). Note separated islet of' hepatocytes (arrow). (Hematoxylin-eosin; original magnification
`x 150.)
`
`Page 5
`
`

`
`HEPATOLOGY Vol. 22, NO. 3, 1995
`
`BANKS ET AL 825
`
`TABLE 4. Histological Changes in 21 Patients
`No. of Patients
`Pattern of Injury
`
`Hepatocellular injury
`Acute
`With spotty necrosis
`With zonal necrosis
`With granulomas
`Chronic
`Chronic hepatitis
`Cirrhosis
`Hepatocellular-cholestatic injury
`Cholestatic injury
`Nonspecific or trivial
`
`9
`2
`6
`1
`6
`5
`1
`1
`2
`3
`
`NOTE. An additional patient showed changes of alcoholic hepati-
`tis and had a history of alcohol use. That case was excluded from
`analysis.
`
`treated mice and that recognition leads to hepatocyte
`injury. These observations suggest that the active metab-
`olite may react with hepatocyte proteins to produce a
`neoantigen. However, the mechanism remains specula-
`tive. The cases of chronic hepatitis may be attributed to
`an immunological mechani~rn.'~
`Factors that appear to affect susceptibility are gen-
`der and disease being treated, although these factors
`may be confounded by the overlap of age, gender, and
`disease in arthritic patients (Table 3). Although most
`of the patients were over 60 years of age, the preponder-
`ance was apparently attributable to disproportionate
`use of the drug in that age group. However, females
`and osteoarthritic patients seemed to have enhanced
`susceptibility. Females outnumbered males by a factor
`of almost four to one, yielding a relative risk of almost
`twice that of males. The relative risk of osteoarthritic
`patients was more than twice that of patients with
`rheumatoid arthritis. The small proportion of patients
`with conditions other than arthritis among those with
`hepatic injury (10%; Table 1) relative to their prepon-
`derant use of the drug (66%; Table 2) presumably re-
`flects shorter duration of use rather than susceptibility
`to injury.
`Examination of the effects of diclofenac treatment on
`serum enzyme levels observed during the premarket-
`ing testing of the drug also suggested that osteoar-
`thritic patients are more susceptible to even minor in-
`jury than patients with rheumatoid arthritis (Table 5).
`Three-fold increases of transaminase levels were sig-
`nificantly more frequent among patients with osteoar-
`thritis, apparently irrespective of age. It was not signif-
`icantly more frequent among females than males.
`These data and the observations among patients with
`overt hepatic injury suggest that susceptibility to mi-
`nor injury is enhanced among patients of either gender
`with osteoarthritis; however, susceptibility to clinically
`significant injury is enhanced further in females.
`The explanation for the seemingly enhanced suscep-
`tibility of patients with osteoarthritis is not apparent.
`It is possible that metabolism of the drug (namely, its
`conversion to a putative toxic metabolite) or its disposal
`
`in rheumatoid arthritis differs from that of osteoarthri-
`tis. However, our data do not permit comparison of
`osteoarthritis with conditions other than rheumatoid
`arthritis, because the small group with other conditions
`was too heterogeneous. The risk may be reduced in
`rheumatoid arthritis, secondary to the effect of the dis-
`ease on drug metabolism. Relevant to this possibility
`is the observation that drug metabolism is reduced in
`animals with Freund's adjuvant-induced arthritis3'
`conceivably inhibiting production of toxic metabolites.
`Alternatively, there may be factors in osteoarthritis
`that enhance toxicity, although the apparently nonsys-
`temic nature of osteoarthritis does not lend itself to
`that argument. The small number of cases in patients
`with other diseases, despite greater use, may be caused
`by shorter duration of use, a greater proportion of
`males, and, possibly, differences in susceptibility.
`The latent period before injury is consistent with that
`of other reports. In most of the cases (85% in this se-
`ries), injury appeared during the first 6 months of ad-
`ministering drug. Particular involvement of elderly pa-
`tients and females has been observed by others.
`However, reports other than the package insert have
`not drawn attention to the apparent particular suscep-
`tibility of patients with osteoarthritis.
`The available data on susceptibility suggest that
`alertness to hepatic injury is particularly appropriate
`in female patients with osteoarthritis. Without clinical
`hallmarks of hypersensitivity to draw attention to an
`adverse reaction, detection of hepatic injury before it
`becomes full blown requires careful attention to subtle
`symptoms. In our view, it suggests the wisdom of moni-
`toring levels of serum enzymes in addition to in-
`structing patients to report fatigue, malaise, anorexia,
`nausea, and/or vomiting. Levels of ALT exceeding
`three-fold the ULN provide reason, in our view, to con-
`sider discontinuing the drug.
`The results of our study do not suggest a simple, fail-
`safe monitoring strategy. Physicians and patients need
`to keep the possibility of hepatic toxicity in mind, if
`and when the patient develops new symptoms or in-
`tercurrent illness, particularly nausea or vomiting. Pe-
`riodic monitoring, as recommended in the package in-
`sert, should take into account the observation that one
`half of the rare fatal cases appeared in the first month,
`half of all cases were detected after the first 2 months,
`and 10% of cases were detected after 9 months.
`The material analyzed and the observations that we
`have derived show the feasibility of analysis of data
`reported to the FDA to characterize the hepatic injury
`produced by a drug. This approach seems particularly
`useful for comparison of hepatic injury occurring after
`a drug has reached the open market with that observed
`during the testing period when frequent routine moni-
`toring leads to earlier detection and discontinuation of
`the drug.
`Comparison of the results of this study with a previ-
`ous one on sulindac-associated injury conducted by the
`same methods3' also shows the inappropriateness of
`referring to NSAID-associated hepatic injury as a uni-
`
`Page 6
`
`

`
`826 BANKS ET AL
`
`HEPATOLOGY September 1995
`
`TABLE 5. Transaminase Levels Greater Than Three Times the Normal Observed During Studies of Diclofenac
`Elevated Serum
`Gender1
`Rheumatoid
`AST and or ALT
`Arthritis N
`Age (yr)
`
`Elevated Serum AST
`
`Osteoarthritis N
`
`P
`
`Females
`<: 65
`-5 65
`Total
`Male
`< 65
`z 65
`Total
`
`759
`136
`895
`
`256
`53
`309
`
`17 (2.244)
`4 (2.94%)
`21 (2.35%)
`
`6 (2.34%)
`0
`6 (1.94%)
`
`NS
`
`NS
`
`2,199
`1,577
`3,776
`
`1,019
`64 1
`1,661
`
`124 (5.644)
`89 (5.64% )
`213 (5.64%)
`
`44 (4.324)
`32 (4.99%)
`76 (4.58% 1
`
`< ,0001”
`
`NS
`< .05*
`
`NOTE. Figures provided by Ciba-Geigy.
`Abbreviation: NS, difference not significant.
`* Difference in prevalence in patient with osteoarthritis vs. rheumatoid arthritis
`
`form phenomenon. Sulindac leads to hepatic injury
`usually accompanied by hallmarks of hypersensitivity
`and most commonly is cholestatic,30 whereas hepatic
`injury associated with diclofenac is uncommonly ac-
`companied by hypersensitivity and is usually hepato-
`cellular. Sulindac hepatic injury is recognized by
`the hypersensitivity features. Biochemical monitoring
`adds little to prevention in contrast to diclofenac injury,
`which might be detected by monitoring while the pa-
`tient is still asymptomatic.
`The prominence of cases of hepatic injury attributed
`to diclofenac among reports submitted to the FDA re-
`ported by Katz and Lovez7 and implicit in the large
`number of cases reported in the literature and analyzed
`in the present report seems at odds with the recent
`report of Jick et al.38 That study of cases of NSAID-
`associated hepatic injury in England recorded hardly
`any cases attributed to diclofenac. The explanation for
`the difference between the report of Jick et alSR and
`the others may be because of differences in the patient
`populations being treated or different durations of use
`of the drug. If use of the drug in the patients whom
`they studied was less than 1 month, injury would not
`be apparent as often as in our study. Only 24% of our
`cases occurred during the first month of use. It is also
`possible that the smaller dose used in about half of the
`group reported by Jick et a1 accounted for fewer cases.
`Furthermore, the apparent relevance to susceptibility
`of gender, and disease being treated suggests that at-
`tempts to reconcile the apparent rarity of diclofenac
`injury noted by Jick et aP8 with the apparent greater
`frequency observed by us require attention to these
`factors.
`Acknowledgment: We are grateful to Drs E. Dennis
`Bashaw, Laurie Burk, Carrie Baum, Thomas Purmitt,
`Ms Jenny Veach, and Ms Dottie Pease for assistance
`with analysis of the data, to Ciba-Geigy for financial
`support and provision of data, and to Mrs Norine Spen-
`cer for preparation of the manuscript.
`REFERENCES
`1. Small HE. Diclofenac sodium. Gastroenterology 1989;8:59.
`2. Dunk AA, Walt KP, Jenkins WJ, Sherlock S. Diclofenac hepati-
`tis. Br Med J 1982;284:1605-1606.
`
`3.
`
`4.
`
`5.
`
`6.
`
`7.
`
`8.
`
`9.
`
`10.
`
`11.
`
`12.
`
`13.
`
`14.
`
`15.
`
`16.
`
`17.
`
`18.
`
`19.
`
`20.
`
`21.
`
`22.
`
`Babany G, Pessayre D, Renhamou J-P. Hcpatite au diclofenac.
`Gastroenterol Clin Biol 1983;7:316.
`Deshayes P, Leloet X, Bercoff E, Fouin-Fortunet H. Hepatite au
`diclofenac. Presse Med 1984; 13:1847.
`Lascar G, Grippon P, Levy V-G. Hepatite aigue mortelle au cours
`d’un traitement par le diclofhac. Gastroenterol Clin Biol
`1984;8:881-882.
`Babany G , Bernuau J , Danan G, Rueff B, Benhamou JP. Hepa-
`tite fulminante chez une femme prenant de la glafenine et du
`diclofenac. Gastroenterol Clin Biol 1985;9:185.
`Breen EG, McNicholl J, Cosgrove E, McCabe J, Stevens FM.
`Fatal hepatitis associated with diclofenac. Gut 1986;27:1390-
`1393.
`Shapiro D, Bassan L, Nahior AM, Scharf Y. Diclofenac-induced
`hepatotoxicity. Postgrad Med J 1986;62:63-65.
`Paret Masana A, Guarga Rojas A, IJrrutai de Diego A, Sabrina
`Leal M. Acute hepatitis and diclofenac sodium ILetter]. Rev Clin
`Esp 1986; 179:476.
`Snijder R], Dinant HJ, Stricker 13. H. CH. Dodelijke lever-
`beschadiging tijdens gebruik van diclofenac. Ned Tijdchr Ge-
`neeskd 1987; 131:2088-2090.
`Lloria G, Larbre J P , Collett PL, Ravault A. Changing the class
`of NSAID in cases of hepatotoxicity. Ann Rheum Dis 1988;47:91-
`93.
`Diggory P, Golding KL, Lancaster R. Kenal and hepatic impair-
`ment in association with diclofenac administration. Postgrad
`Mcd J 1989;65:507-508.
`Hovette P, Touze J E , Debonne JM, Delmarre B, Kogier C,
`Schmoor P, Aubry P. Cholestatic hepatitis and acute kidney in-
`sufficiency during treatment with diclofenac. Ann Gastroenterol
`Hepatol 1989;25:257-258.
`Iveson TJ, Ryley NG, Kelly PMA, Trowell JM, McGee OD, Chap-
`man RWG. Diclofenac-associated hepatitis. cJ Hcpatol 190; 10:85-
`90.
`Helfgott SM, Sandberg-Cook J, Zakim D, Nestler *J. Diclofenac-
`associated hepatotoxicity. JAMA 1990; 264:2660-2662.
`Sallie RW, McKenzie T, Rced WD, Quinlan MF, Shilkin KB.
`Diclofenac hepatitis. Aust N Z J Med 1991;21:251-252.
`Ouellette GS, Slitzky BE, Gates JA, Lagarde S , West AR. Revers-
`ible hepatitis associated with diclofenac. J Clin Gastroenterol
`1991; 13:205-210.
`Purcell P, Henry D, Melville G. Diclofenac hepatitis. Gut
`1991; 32: 1381 -1385.
`Nos P, Palau J , Perez-Aguilar F, Rerenguer J . Hepatitis associ-
`