3.5.2 Clinical Outcomes
Viral reactivation was shown to be significantly associated with higher risk of outcomes of mortality in 1 year (OR, 3.9; 95% CI, 1.2-12.2; p=0.02), dialysis initiation (OR, 3.4; 95% CI, 1.1-10.9; p=0.04), increased length of hospital stay (IRR, 2.2; 95% CI, 1.4-3.3; p<0.01), and recurrent flares in 1 year among DRESS patients (OR, 3.1; 95% CI, 1.2-8.2; p=0.02) (Table 4). On sub-analysis with respect to individual viruses, the association with mortality, ICU stay, need for dialysis and length of stay (LOS) was significantly associated with CMV reactivation. Similarly, when there were 2 or more viral reactivations detected, the risk of inpatient mortality (OR, 5.8; 95% CI, 1.7-20.7; p<0.01) as well as 1 year mortality was significantly higher. (OR, 10.0, 95% CI, 2.9-34.9; p<0.01).

3.6 Impact of Anti-viral treatment in patients with viral reactivation.

To explore the role of antiviral treatment in DRESS, a preliminary comparative analysis was performed. Among the 39 patients with reactivation, 13 received anti-viral therapy: 11 had serological positive viral PCR, whereas 2 others had additional evidence of pathology-proven clinical disease (CMV pneumonitis, CMV hepatitis). Antivirals included ganciclovir (n=7,54%), acyclovir (n=5, 38%) and valganciclovir (n=1,8%). Treatment decisions were made by the primary physicians. There were no significant differences in outcomes (Mortality, ICU, need for dialysis, LOS) between both groups
4. Discussion
This current study examines the impact of viral reactivation in a large cohort of patients with DRESS syndrome. There were a few unique observations. Firstly, herpes viral reactivation is a common but non-universal phenomenon, occurring in 42% of patients. Secondly, viral reactivation may not be detected at the onset of the rash (Figure 1) and 28% of patients with reactivation were detected on serial testing. Thirdly, there were no baseline clinical, causative drug or treatment factors that predicted reactivation. Lastly, our study suggests the association of viral reactivation with a certain clinical phenotype – such patients were more likely to have renal involvement as well as poorer outcomes, namely recurrent flares, need for dialysis, length of stay and death.
Identification of predictors of severe, life-threatening DRESS remains a research gap. In a series of 15 patients with ICU stay and death19, HHV6 reactivation was present in 6/7 tested (No other herpes viruses were tested). Similarly, CMV reactivation has been anecdotally reported with severe outcomes. Our current study further clarifies this association. HHV6 in itself was not associated with ICU stay or death. However, when it is associated with multiple viral reactivation and / or CMV reactivation, poorer prognosis exists.
There were other significant negative findings that are of note. Firstly, there were no baseline factors which predicted viral reactivation. This proved that reactivation occurred independent of systemic steroid treatment, type of culprit drug or baseline factors. Nevertheless, valproic acid has been reported to increase replications of HHV6,20 and CMV.21 Secondly, in our exploratory analysis of antiviral treatment in a small cohort of viral-reactivated DRESS cases, there does not appear to be significant difference in outcomes. This warrants further evaluation in a systematic and controlled protocol.
The pathophysiology of DRESS and the role that herpes viruses play remains poorly defined. Although viral reactivation is associated with poorer outcomes, the cause versus effect conundrum remains unresolved. A few possibilities exist: 1) Viral reactivation exist as a bystander effect due to immunodysregulation – not dissimilar to an immune reconstitution phenomenon in graft versus host disease or in critically ill/immunosuppressed subjects. 2) Viral reactivation results in the initiation of the drug allergy. 3) Direct reactivation of viruses from drug / drug metabolites. 4) A combination of the above – with a primary drug-specific immune response as the initiating event, then viral reactivation in certain individuals with an associated secondary anti-viral/self-response. The lack of universal reactivation in all DRESS patients and reactivation occurring after the onset of DRESS symptoms argues against it being the initial trigger or a necessity for DRESS to occur. Furthermore, in-vitro and in-vivo evidence such as the identification of drug-specific T cells in Allopurinol SCARs and positivity on patch testing to the culprit drug in DRESS patients supports the argument for a primary drug-mediated response.11
There were various limitations in our study. These included the retrospective design with its inherent flaws. Although viral studies were performed in the acute phase of the disease, the timing of these studies depended on various factors including time to hospital admission and/or dermatological referral. This may have impacted on the proportion of reactivation as well as the latency between onset of symptoms and viral reactivation. This was partially ameliorated by repeated sampling which was performed in 24% of our patients. In our cohort of 93 patients, 28% had HHV6 reactivation and these results are comparable to other validated DRESS cohorts utilizing quantitative PCR viral studies. In those cohorts, HHV6 reactivation varies between 39%-45%.13, 24 Moreover, the latency between symptoms and reactivation in our cohort is similar to published series.13, 23 Although we showed univariate analysis for each outcome e.g. mortality, dialysis etc, these results may be limited by the small sample size. A multivariable model for any outcome was not possible due to the small sample and missing values. Our analysis on anti-viral treatment, though novel, is not conclusive due to the non-controlled nature, small cohort and risk of treatment bias. Nevertheless, these preliminary data warrants further evaluation.
There were certain strengths in this study. This was a large cohort of DRESS patients that was validated against standardized criteria. Dermatological care was provided by the same clinical team within the same care setting. This would reduce the bias introduced by care variability and the “centre” effect.
In conclusion, our study suggests that herpes viral reactivation, when present identifies patients at risk of poorer outcomes. Our work is unable to prove a causal or pathogenic association and further work is needed to understand the role of virus reactivation in DRESS disease mechanism, identify patients at risk of reactivation as well as potential impact of anti-viral treatment.