Search results
A total of 1099 publications were identified during the initial search. 444 duplicates were removed (Fig. 1). After screening through the title and abstracts of 655 publications, 113 publications (18 meta-analyses, 93 case-controlled studies and 2 case series) were identified to be meeting the inclusion and exclusion criteria for full-text review. After full text review, 32 articles were excluded. Further hand searching of journals led to an addition of 2 case-controlled studies and a total of 83 studies (13 meta-analyses and 70 case-controlled studies) were included in the umbrella review.
Study characteristics
The selected articles were published between 2008 to 2023. A total of 83 studies were included in this umbrella review. Each antiepileptic was analyzed separately. For Carbamazepine, there were 42 case-control studies(11–14)(15–52)and 7 meta analyses(39, 53–58). For Lamotrigine, there were 30 case-control studies(13, 15, 17, 26, 34, 39, 42, 46, 49, 59–72) and 3 meta-analyses(66, 68, 70). As all the case-control studies for Carbamazepine and Lamotrigine were included in the meta-analyses, we only analysed the results of the meta-analyses.
For Oxcarbazepine (OXC), there were 3 case-control studies(20, 29, 73) and 2 meta-analyses(57, 74). For Phenytoin (PHT), there were 9 case-control studies(18, 20, 27, 49, 75–79) and 3 meta-analyses(68, 80, 81). For Phenobarbital (PB), there were 4 case-control studies(15, 29, 49, 82) and 1 meta-analysis(81). As the recent case-control studies for the Oxcarbazepine, Phenytoin and Phenobarbital were not represented in their recent meta-analyses, we analysed the case-control studies separately. The quality of meta-analyses was appraised with AMSTAR-2 and the quality of case control studies were appraised with NOS as shown in Table 1 and Table 2 respectively under “Supporting Information”.
Carbamazepine (CBZ) and HLA-B*15:02
A total of 7 relevant meta-analyses were identified(39, 53–56, 58, 83). 4 of these studies were found to be of critically low quality. However, the most recent and largest studies(81, 84) were rated as high in quality(10). (See Table 1 under “Supporting Information”).
All 7 meta-analyses consistently demonstrated a strong association between HLA-B*15:02 and Carbamazepine -induced SJS/TEN, with odds ratios ranging from 19.1(95%CI:7.9–46.1) to 164.5(95%CI:73.6-367.8). These results are visually depicted in Fig. 2. Notably, the study conducted by Biswas et al(84) is the most recent and comprehensive among these meta-analyses. It encompassed 38 studies involving of 1346 cases, 2504 tolerant controls and 564 population controls. Based on this analysis, the odds ratio for developing Carbamazepine -induced SJS/TEN in the context of HLA-B*15:02 is 26.3(95%CI:21.4–32.4).
The current literature also indicates that the association between HLA-B*15:02 and Carbamazepine -induced SJS/TEN extends beyond the Asian population. In an analysis of 36 studies, including 34 studies involving Asian patients and 2 studies involving Caucasian patients, Biswas et al(84) demonstrated that the odds ratio (OR) of developing SJS/TEN is also elevated in Caucasian [OR14.8(95%CI:9.0-24.6), compared to Asians compared OR 11.65 (95%CI: 1.68–80.7)].
Lamotrigine (LTG) and HLA-B*15:02
A total of 3 meta-analyses(66, 68, 70) examined the relationship between HLA-B*15:02 and Lamotrigine-induced SJS/TEN. The studies were published between 2015 and 2018 and involved exclusively Asian patients. No studies relevant to this question have been added to the literature since the publication by Deng et al in 2018(70). Of these studies, two were assessed to be of low quality(66, 68) and one was assessed to be of moderate quality(70). (See Table 1 under “Supporting Information”).
All 3 studies consistently demonstrated a strong association between HLA-B*1502 and Lamotrigine-induced SJS/TEN, with odd ratios ranging from 2.55(95%CI:1.29–5.04) to 4.83(95%CI:1.27–18.45) as visually depicted in Fig. 3. The study conducted by Deng et al[ref] in 2018 reported the largest sample size. In this study which included the data from 7 studies comprising of 54 cases of Lamotrigine-induced SJS/TEN and 313 controls, a significant association between HLA-B*15:02 and Lamotrigine-induced SJS/TEN was observed [OR of 2.55 95% CI: 1.29–5.04)].
Oxcarbazepine and HLA-B*15:02
2 meta-analyses(57, 74) were identified for Oxcarbazepine as shown in Fig. 4. In 2018, Liu et al pooled the data from three studies, involving 154 patients, and found a positive association between HLA-B*15:02 and severe cutaneous adverse reactions(SCAR) induced by Oxcarbazepine [OR 18.13 (95%CI:6.77–48.56)]. In the same year, Tangamornsuksan also reported a positive association between HLA-B*15:02 and Oxcarbazepine-induced SJS/TEN based on two studies, involving 142 patients [OR30.2(95%CI:3.45–264)].
As the study by Tangamornsuksan et al examined SJS/TEN (as opposed to Liu which also examined SCAR which includes not only SJS/TEN but also drug reaction with eosinophilia and systemic symptoms (DRESS syndrome)) but was found to have missed one study, we performed a repeat meta-analysis using the findings of 3 case-controlled studies involving 346 subjects.(20, 29, 73) In this analysis, a strong positive association was again found between HLA-B*15:02 and Oxcarbazepine-induced SJS/TEN [OR29.23 (95% CI 10.71–79.79)].
Phenytoin and HLA-B*15:02
3 meta-analyses(68, 80, 81) investigated the association between HLA-B*15:02 and Phenytoin-induced SJS/TEN as visually depicted in Fig. 5A. All 3 studies found a positive association between HLA-B*15:02 and Phenytoin induced SJS/TEN (Odd ratio of 1.99 (95% CI:1.28–3.10) and 3.15 (95%CI:1.67–5.94). The most recently published meta-analysis by Rashid et al 2022(81) included a study which examined the association of all Severe Adverse Cutaneous Reaction (SCAR) and not SJS/TEN alone, explaining the lower odds ratio compared to the other 2 meta-analyses. As this may potentially affect the result of the study, we repeated the meta-analysis from 13 case-control studies(15, 18, 20, 27, 59, 72, 75–79, 85, 86) from our database. All studies show a significant association found between HLA-B*15:02 and SJS/TEN in patients on Phenytoin (OR of 3.00, 95% CI 1.65–2.47, I2 64%, p < 0.01) as depicted in Fig. 5B. All the 13 studies were on Asian populations, including Thais, Han Chinese, Malays, and Indians.
Notably, the subgroup analysis found a significant association only in Han Chinese patients. Analysis did not show evidence of publication bias and pooled OR was not significantly affected by the omission of any individual study.
Phenobarbital and HLA-B*15:02
The study identified one meta-analysis [54] which examined the relationship between HLA-B*15:02 and SJS/TEN. However, as this study examined all cutaneous adverse reactions, the decision was made to repeat the meta-analysis.
This meta-analysis included 3 case-controlled studies involving a sample size of 67 patients and did not show an association between HLA-B*15:02 and phenobarbital-induced SJS/TEN [OR 2.89(95%CI:0.36–22.97)] as depicted in Fig. 6.