Cumulative incidence of overall laboratory abnormalities prior to and during isotretinoin therapy. Abnormalities are defined as results above the normal range for triglyceride, total cholesterol (TC), and transaminases (aspartate aminotransferase [AST] values were used if alanine aminotransferase [ALT] values were not available) and as below the normal range for white blood cells (WBCs), hemoglobin, and platelets. Error bars represent upper bound of 95% confidence interval.
Zane LT, Leyden WA, Marqueling AL, Manos MM. A Population-Based Analysis of Laboratory Abnormalities During Isotretinoin Therapy for Acne Vulgaris. Arch Dermatol. 2006;142(8):1016-1022. doi:10.1001/archderm.142.8.1016
To determine the incidence of abnormal laboratory test results among isotretinoin users.
Comprehensive managed care health plan in Northern California.
The study population comprised 13 772 patients aged 13 to 50 years with acne, undergoing oral isotretinoin therapy between March 1995 and September 2002.
Main Outcome Measures
Laboratory values for serum triglyceride, total cholesterol, and liver transaminase levels; white blood cell count, hemoglobin level, and platelet count; and frequency of abnormal laboratory results by severity grade as defined by the National Cancer Institute Common Terminology Criteria for Adverse Events v3.0.
Substantial increases in the cumulative incidence of abnormalities were seen in serum lipid and transaminase levels, but not in hematologic parameters, during isotretinoin treatment compared with the baseline period. The cumulative incidence of new abnormalities in patients with normal values at baseline was 44% for triglyceride level, 31% for total cholesterol level, and 11% for transaminase level. Moderate to severe abnormalities in lipid and transaminase levels were generally transient and reversible. New abnormalities in hematological test results were uncommon.
The incidence of abnormally high serum lipid levels during isotretinoin treatment may be greater than previously estimated. Elevations in transaminase level are generally mild. Normal baseline values of serum lipid and transaminase levels do not preclude the development of new abnormalities during isotretinoin treatment. Routine monitoring of white blood cell count, hemoglobin level, and platelet count during isotretinoin therapy may be of little utility without clinical suspicion of an abnormality. The clinical significance of laboratory abnormalities during isotretinoin therapy remains to be determined.
Oral isotretinoin is currently the most effective acne treatment available, with reported long-term remission rates as high as 89%.1 Case reports of such systemic complications as pancreatitis,2- 4 drug-induced hepatotoxicity,5 leukopenia,6 and thrombocytopenia7 have prompted the widespread practice of frequent laboratory monitoring of patients undergoing isotretinoin therapy. However, 2 retrospective studies found little evidence for the use of routine laboratory testing.8,9
To our knowledge, the incidence of laboratory abnormalities during isotretinoin therapy has not been described in large, population-based samples. The Accutane (Roche brand of isotretinoin; Nutley, NJ) package insert cites marked elevations in serum triglyceride level in approximately 25% of patients and mild to moderate elevations of liver enzyme level in 15% of patients receiving Accutane in clinical trials.10 Other studies reporting the frequency of laboratory abnormalities have identified elevated triglyceride level in 5% to 18% of subjects and elevated total cholesterol level in 6% to 32%.8,9,11,12 Limited sample sizes diminish the precision of these studies' estimates, and the lack of standard criteria for defining abnormal values among these studies makes them difficult to compare. We report herein the incidence of laboratory abnormalities using standardized grading criteria in a large, population-based cohort of 13 772 patients with acne treated with isotretinoin between 1995 and 2002.
The Kaiser Permanente Medical Care Program, Northern California (KPNC), a group practice nonprofit integrated health plan, provides comprehensive medical services primarily in the urban areas around the greater San Francisco Bay and Sacramento metropolitan areas, serving approximately 3.3 million members (roughly 30% of the population in these areas). The health plan membership is ethnically and socioeconomically heterogeneous and generally representative of the San Francisco Bay Area population, although the lowest and highest strata of income are somewhat underrepresented.13 The KPNC maintains comprehensive databases of such medical information on its membership as outpatient and inpatient encounters and diagnoses, pharmacy prescription records, and records of laboratory tests and diagnostic procedures.
The study cohort consisted of KPNC members between the ages of 13 and 50 years, with at least 1 coded or write-in outpatient diagnosis consistent with acne vulgaris, and whose pharmacy record indicated that they filled a prescription for at least 4 weeks of isotretinoin therapy after March 1, 1995, and had undergone at least 1 course of therapy by September 30, 2002. For the 16.5% of the participants who had undergone more than 1 course of isotretinoin therapy during this period, only the first course of isotretinoin therapy was considered. A course was defined as continuous filling of isotretinoin prescriptions with no more than a 60-day interruption. Potential participants were excluded if (1) they were not fully enrolled in KPNC from 2 months prior to starting isotretinoin therapy through 3 months after the first course of treatment or (2) their first isotretinoin prescription during the study period was written by a health care provider outside the KPNC system.
The Kaiser Foundation Research Institute institutional review board and the University of California, San Francisco, Committee on Human Research granted approval for this study. We obtained all data for this study from KPNC databases.
The primary predictors in this study were the periods of isotretinoin exposure and nonexposure. Three periods were considered: the “baseline” period began 2 months prior to the first prescription of isotretinoin and ended 2 weeks following the start date of therapy to allow for the practice of baseline laboratory testing concurrent with the first prescription. The “treatment” period extended from the end of the baseline period through the completion of therapy as predicted by pharmacy records. The 3-month period following the treatment period was labeled the “posttreatment” period.
Results were ascertained for the following laboratory tests during the study period: triglyceride, total cholesterol, alanine aminotransferase (ALT), and aspartate aminotransferase (AST) levels; white blood cell (WBC) count; hemoglobin level; and platelet count. To accommodate differences in rates of testing across the cohort, all test results for each patient from each exposure period were collectively examined, and only the highest value for each test (lowest value for WBC count, hemoglobin level, and platelet count) from each period was considered. The National Cancer Institute (NCI) Common Terminology Criteria for Adverse Events v3.0 (CTCAE)14 were used to delineate the grades of laboratory abnormality (Table 1). The CTCAE is a well-accepted system for grading the adverse effects associated with treatment modalities and allows for standardized reporting across groups and modalities without regard to chronicity. The CTCAE does not maintain a separate grading system for pediatric populations. Where the CTCAE grades were defined relative to upper and lower limits of normal, the corresponding reference values for adults from the KPNC laboratory were applied. Since the CTCAE does not specifically grade elevations in WBC count, hemoglobin level, or platelet count, laboratory values above the upper limit of normal were graded as “high.”
Since KPNC guidelines recommend against obtaining ALT and AST values simultaneously owing to redundancy between these tests and because ALT level is a more specific marker of hepatocellular injury, we primarily examined ALT results and considered AST values only when ALT results were not available. Herein, this value will be referred to as “transaminase.”
The incidence of laboratory testing during each exposure period was expressed as the number of patients receiving at least 1 test divided by the total number of patients in the full cohort. Cumulative incidence of any laboratory abnormality during the baseline and treatment periods was expressed as the number of patients demonstrating a laboratory test result outside the normal range during a given period divided by the total number of patients receiving that test during the same period. Cumulative incidence of a new laboratory abnormality was expressed as the proportion of patients with at least 1 abnormal laboratory value during the treatment period among those with a normal range test result during the baseline period. Confidence intervals (95%) were calculated for each incidence estimate. All statistical analyses were carried out using Intercooled Stata 8 (version 8.2; StataCorp, College Station, Tex).
The median age of the cohort (N = 13 772) was 19 years, and 51% were male. Nearly all patients (98%) had records of 2 or more diagnoses of acne vulgaris, while 2% had only 1 acne diagnosis. The number of subjects initiating isotretinoin treatment remained stable across the years of the study, with slight declines in the first and last years due to exclusions for insufficient exposure period. The median duration of the treatment course under study was 21 weeks, with a median cumulative dose of 9 g and median calculated daily dosage of 65 mg. Patient weight data during the treatment period were not available with enough consistency to allow the calculation of daily dose in milligrams per kilograms per day. The majority (89%) of initial isotretinoin prescriptions were written by practitioners in dermatology.
Overall, the proportion of patients tested was high during the baseline and treatment periods. Among the tests, triglyceride and transaminase levels were measured in the greatest proportion of subjects during the baseline (both 90%) and treatment (88% and 85%, respectively) periods. The remaining tests under investigation were performed in at least 78% of patients at baseline and at least 66% of patients during treatment. Rates of testing declined substantially in the posttreatment period, with less than 20% of the cohort receiving each test.
The Figure shows the cumulative incidence of abnormalities for each laboratory parameter among those tested during the baseline and/or treatment period. All tests showed a statistically significant increase in the incidence of abnormal values during treatment compared with baseline, with the exception of platelet count. The incidence of abnormally elevated serum triglyceride level increased the most from the baseline to the treatment period (14% vs 50%; P<.001), followed by total cholesterol (15% vs 41%; P<.001) and transaminase (5% vs 13%; P<.001) levels. The incidence of abnormally low values among hematological parameters did not differ substantially between the baseline and treatment periods. However, small but statistically significant differences between baseline and treatment incidence of abnormally low hemoglobin level (2.5% vs 4.1%; P<.001) and WBC count (1.3% vs 2.7%; P<.001) were observed.
The abnormal laboratory results classified by grade of abnormality are given in Table 2. Compared with baseline values, abnormal triglyceride values were more severely abnormal during treatment. While the majority of these abnormalities during treatment were in the grade 1 category, more than 7% of patients tested demonstrated a moderate to severe (grade 2 or higher) abnormality, with 9 patients (0.07%) exhibiting triglyceride levels above 1500 mg/dL (17.0 mmol/L) (grade 4). Similar shifts were seen in total cholesterol and transaminase values, with the proportion of patients with a moderate to severe abnormality increasing from 0.2% at baseline to 1.4% during treatment (P<.001) for total cholesterol level and from 0.4% to 1.5% (P<.001), respectively, for transaminase level.
Moderate to severe abnormalities in triglyceride, total cholesterol, and transaminase levels were generally transient and reversible. Among those subjects with such abnormalities who received posttreatment testing, the proportion returning to normal or grade 1 levels by the end of the posttreatment period was 92% for transaminase level, 80% for triglyceride level, and 79% for total cholesterol level.
Abnormalities in WBC count, hemoglobin level, and platelet count were generally mild and infrequent. A slight increase in the incidence of moderate to severe abnormalities was seen in WBC counts during treatment compared with baseline (0.3% vs 0.6%; P = .001), while no significant difference was seen in the proportion of moderate to severe abnormalities for hemoglobin level (P = .50). The incidence of platelet count abnormalities did not differ between treatment and baseline periods in any of the grades. One apparently spurious case of a grade 3 abnormality in platelet count was identified. This low platelet count was the only abnormal value among 6 recorded for that patient during the treatment period, including 1 result in the normal range 3 days after the low value was measured.
The cumulative incidence of new abnormal test results during the treatment period in patients who tested within the normal range at baseline is given in Table 3. Elevations in triglyceride level showed the highest incidence, with 44% of patients demonstrating a value above the normal range during the treatment period. Moderate to severe abnormalities were seen in 4%. While a substantial incidence of newly elevated serum cholesterol levels was also found (31%), only 9 cases (0.1%) exceeded grade 1. Cumulative incidence of new transaminase abnormality during the treatment period was 11%, with 1% being grade 2 or higher.
Among these patients with normal test results in the baseline period, incident moderate to severe abnormalities were also transient and reversible. Among those subjects with such abnormalities who received subsequent testing, the proportion returning to normal or grade 1 levels by the end of the posttreatment period was 93% for transaminase level, 90% for triglyceride level, and 100% for total cholesterol level.
New abnormalities in hematological laboratory test results among those testing within the normal range at baseline were very uncommon. Among WBC, hemoglobin, and platelet tests, the only incident case of abnormality beyond the grade 2 level was a grade 3 platelet count abnormality that was likely to be spurious (described in the previous subsection).
We describe in a large population of patients with acne the incidence of laboratory abnormalities that occur during typical management with oral isotretinoin. While substantial increases in more severe abnormalities in serum lipid and transaminase levels were observed during the treatment period compared with the baseline period, the cumulative incidence of hematologic abnormalities showed little or no change. Abnormalities in serum lipid levels were common during isotretinoin therapy, while abnormalities in transaminase level were less common and generally mild. Moderate to severe elevations in serum lipid and transaminase levels were transient in most (>79%) subjects. Normal baseline values for serum lipid or transaminase levels did not preclude the development of an abnormality during treatment. The clinical significance of laboratory abnormalities during isotretinoin therapy remains to be elucidated.
The cumulative incidence of abnormally elevated serum lipid levels during isotretinoin therapy may be higher than some previously published estimates. While we applied the criteria defined by the NCI CTCAE, we have also analyzed our data using laboratory definitions from prior studies to allow for direct comparison. Results of 4 studies examining abnormalities in serum lipid levels during isotretinoin therapy are summarized in Table 4, compared with data from the present study.
The greatest increase in cumulative incidence of abnormalities from baseline to treatment was observed in serum triglyceride level. Half of treated patients, including 44% of patients who tested within the normal range at baseline, had triglyceride levels above the normal range (>150 mg/dL [>1.7 mmol/L]) during treatment. Similar changes in serum total cholesterol level were also seen during treatment. About 40% of patients had total cholesterol levels above 200 mg/dL (>5.2 mmol/L) during isotretinoin therapy, and 30% of patients with normal baseline levels developed abnormal elevations during treatment.
Few studies have examined the incidence of new abnormalities in serum lipid levels among patients with normal levels at baseline. Bershad and colleagues11 reported elevations in total cholesterol level to 240 to 300 mg/dL (6.2-7.8 mmol/L) in 5.8% of patients whose levels were normal at baseline, an incidence criterion met by 6.0% of patients in the present study.
While all triglyceride and total cholesterol samples were assumed to have been drawn from fasting patients, some measurements may have been affected by inadequate adherence to fasting instructions. In addition, differences in dietary practices may contribute to the discrepancies in serum lipid level abnormalities seen between our study and those conducted in other countries.8,9,12
Overall, the cumulative incidence of new abnormalities in liver transaminase level was low (11%), and most elevations were mild (91%). Methodological differences between the present study and previous studies examining transaminase level abnormalities during isotretinoin therapy4,9,12 preclude direct comparison of incidence estimates. Abnormalities in transaminase level may also be related to comorbid hepatic conditions, concomitant medical therapy, or alcohol consumption, which were not measured in the present study.
While there have been published case reports of dramatic reductions in hematological parameters in the setting of isotretinoin use,6,7,16 our results reveal very few abnormalities during treatment. We conclude that except for patients in whom a hematologic abnormality is suspected prior to or during treatment, the diagnostic yield of routine monitoring of WBC count, hemoglobin level, and platelet count during isotretinoin therapy is too low to be clinically useful.
Standardized grading scales have been considered prerequisites for the improvement of adverse event reporting in clinical trials.17 There is no established grading system for laboratory abnormalities specifically for dermatologic treatments. The use of different arbitrary cutpoints by individual studies precludes direct comparisons. Scales containing multiple ordinal categories are more informative than the single cutpoints used in some studies. Of the 3 well-known standardized scales for grading laboratory abnormalities, we consider the NCI CTCAE the most complete and relevant for the evaluation of isotretinoin therapy, given the additional role of isotretinoin as a chemotherapeutic18,19 or chemopreventive19 agent. Use of the CTCAE scales here is not intended to ascribe specific risk of adverse clinical outcomes, but to provide a standardized and comprehensive system for describing the distribution of laboratory abnormalities during periods of isotretinoin exposure and nonexposure.
The population-based approach of this study provides increased statistical power; greater precision of estimates; improved generalizability of results; and appropriate perspective on rare extreme cases. This retrospective analysis of medical record data yields results that reflect what is observed in the typical clinical management of isotretinoin-treated patients, unrestricted by the selection criteria or procedural protocol of clinical trials. Furthermore, it incorporates the diversity of prescribing practices used by a multitude of practitioners who may have different thresholds for risk, preferred dosing regimens, and responses to laboratory abnormalities.
There are limitations to this study that deserve mention. While the unavailability of consistent patient weight data in the medical record precluded the assessment of weight-adjusted dosing, the average calculated dose for the cohort was 65 mg/d, which falls within the common dosing range of 0.5 to 1.0 mg/kg per day in a population with an average weight near 70 kg. Furthermore, in a retrospective analysis in which laboratory testing frequency is not set a priori, interpretations are limited by a potential testing bias. Namely, those patients with abnormal values may be more likely to receive subsequent testing than those with normal values, while those patients with normal test results may be less likely to receive additional testing that could reveal future abnormalities. Such a bias could influence the results toward underestimating the frequency of laboratory abnormalities among those patients with normal values at baseline and potentially overestimating the incidence and severity of abnormality by repeatedly testing patients with a prior abnormality.
Our study did not examine adverse clinical outcomes and thus cannot estimate the ability of abnormal laboratory results to predict such outcomes. In clinical practice, laboratory abnormalities should be evaluated in the clinical context of the individual patient. Neither does the presence of a laboratory abnormality necessarily signal the presence of an adverse clinical outcome, nor does the absence of a laboratory abnormality preclude the possibility of an adverse clinical outcome. Patients with acne who develop large increases in triglyceride levels during isotretinoin therapy are at increased risk for future hyperlipidemia and the metabolic syndrome.20,21 We plan to examine the incidence of adverse clinical outcomes in a future analysis.
Correspondence: Lee T. Zane, MD, MAS, Department of Dermatology, University of California, San Francisco, 1701 Divisadero St, Third Floor, San Francisco, CA 94143-0316 (ZaneL@derm.ucsf.edu).
Financial Disclosure: None reported.
Accepted for Publication: May 23, 2006.
Author Contributions:Study concept and design: Zane, Leyden, and Manos. Acquisition of data: Zane, Leyden, and Manos. Analysis and interpretation of data: Zane, Leyden, Marqueling, and Manos. Drafting of the manuscript: Zane and Marqueling. Critical revision of the manuscript for important intellectual content: Zane, Leyden, Marqueling, and Manos. Statistical analysis: Zane, Leyden, and Manos. Obtained funding: Zane. Administrative, technical, and material support: Zane, Leyden, Marqueling, and Manos. Study supervision: Zane and Manos.
Funding/Support: This study was supported by the Department of Dermatology, University of California, San Francisco, and a Dermatology Foundation Dermatologist Investigator Research Fellowship (Dr Zane). This study received no funding from any pharmaceutical company.
Acknowledgment: We thank Thomas Newman, MD, MPH, for his critical assessment of this manuscript.