This study is a large population-based matched-control prospective follow-up survey on a Taiwanese population to explore the association between herpes zoster and the subsequent risk of lymphoid malignancies. To the best of our knowledge, this is currently the largest matched-control study of an Asian population. Our results showed that preceding herpes zoster infection is an independent risk marker for subsequent lymphoid malignancies in Taiwanese subjects from univariate and multivariate analyses. There are several retrospective case–control studies that have explored the risks of preceding infections and subsequent lymphoid malignancies in Western populations
[13–17]. Among the preceding infections, herpes zoster was shown to be associated with both Hodgkin’s disease and non-Hodgkin’s lymphoma in a hospital-based case–control study in Italy
 and an increased risk of CLL in male U.S. veterans
. In a large case–control study using the U.S. SEER database, herpes zoster showed an increased risk of CLL and the risk increased with increasing severity or frequency of herpes zoster
. An increased risk of multiple myeloma or monoclonal gammopathy of undetermined significance was found in patients with history of herpes zoster in Italian and in white and black male U.S. veterans
[15, 16]. Similarly an increased risk of lymphoplasmacytic lymphoma or Waldenstrom macroglobulinemia was found from Swedish population-based registries
The association of herpes zoster and subsequent risk of internal malignancies has been reviewed for decades, however most studies have not found an association between herpes zoster and an increased risk for the development of future malignancies
[11, 12]. In our study, the association of herpes zoster and risk of cancer was seen in lymphoid malignancies but not in non-lymphoid malignancies, which is compatible with other studies and suggests that herpes zoster is an independent risk marker for lymphoid malignancies but not a risk marker for other cancers.
Some host factors have been found to be associated with the development of lymphoma including inherited genetic factors, infections, autoimmune diseases, and environmental exposure
. HIV-AIDS is a known risk factor for lymphoid malignancies
[1–4], and patients with HIV infection and other known viral infections that could be coded from the ICD-9-CM were excluded in our analysis to exclude the influence of viruses other than herpes zoster. Autoimmune diseases and chronic inflammatory disorders have been shown to be associated with lymphoid malignancies
[27–29]. Although the proportion of autoimmune diseases in patients was higher than in the controls (4.2% vs. 2.9%), the covariates of CCI including rheumatologic diseases were adjusted in Cox proportional hazard regressions, and a relatively small difference was noted in adjusted HRs from step-wise removal of comorbidities, indicating that the influence of autoimmune diseases is low. Similarly, other systemic comorbidities may play a role in the association with the development of lymphoid malignancies. The adjustment of covariates of CCI, and a relatively small difference in adjusted HRs from step-wise removal of comorbidities, are helpful to diminish the confounding factors caused by systemic comorbidities.
The total case number of herpes zoster (n = 42,498) in our cohort before diagnosis of any malignancies is comparable with some studies focusing on herpes zoster using Taiwan NHIRD
[30, 31], which showed the incidence rate of herpes zoster in Taiwan is estimated around 4.89 to 4.97 cases per 1000 person-years, indicating the possibility of underdetection in our cohort is low. In addition, our dataset contained inpatient and outpatient compartments, preventing the selection and information bias common in hospital-based studies. Moreover, most of the published study designs have been retrospective case–control studies. Prospective study that shows infection preceded lymphoid malignancies and that infection is associated with subsequently elevated lymphoid malignancies risk can be especially compelling.
The distribution of lymphoid malignancies in Asian populations is different from those in Western countries
[18–22]. The incidence of low grade lymphoma, CLL, and Hodgkin’s disease is lower than in Western populations, and the incidence of T cell lymphoma is higher, indicating racial or geographical differences in the etiology of lymphomagenesis. Although the cause is unclear, searching for an antecedent exposure history prior to lymphoid malignancies is helpful in determining the causes of the geographical differences. Furthermore, to better clarify the role of antecedent herpes zoster infection with subsequent lymphoid malignancies, it is important to perform analysis in a different population. Our study provides evidence supporting the association between herpes zoster infection and the subsequent development of lymphoid malignancies in Taiwanese subjects, and warrants further studies to explore the mechanism between herpes zoster and lymphomagenesis.
Many studies have attempted to explore the mechanisms between viral infections and lymphomagenesis. The potential mechanisms by which most viruses play a role in lymphomagenesis are by introducing alternations in the genome through the incorporation of viral genes, or viral infection itself may induce a state of chronic inflammation
. Some viral infections, like EBV, can directly infect and transform lymphocytes, disrupt normal cell function, and promote cell division
. The HIV infection is uinque in causing depletion of CD4+ T-cells and induces profound cell-mediated immunodeficiency which permits dysregulated proliferation of B lymphocytes. Some viral infections, like HCV, may induce chronic immune stimulation and persistent lymphocytes activation
. Although the mechanism between VZV and lymphomagenesis is currently unknown, the fact that the VZV establishes latency after primary infection, and herpes zoster results from a reactivation of latent VZV infection which is often associated with a decline in cell-mediated immunity
[8, 9] imply the possibility of the role of chronic immune stimulation. At the same time, since the decline of VZV-specific cell-mediated immunity may be a consequence of immunosuppression
, an alternative explanation is that the herpes zoster infection is a marker of immune suppression, and such an immune suppressed condition is a potential cause of lymphoid malignancies thus herpes zoster infection is a risk marker for subsequent lymphoid malignancies. Further studies are needed to determine if VZV may trigger continuous antigenic and immune stimulation, or a disturbed immune function, and how these may be associated with the development of lymphoid malignancies.
There are limitations to our study. First, the ICD-9-CM coding system is not sufficient to classify the subtypes of lymphoid malignancies according to the 2008 WHO classification and the limited events during follow-up (n = 48), leading to difficulty in performing histological subtype analysis, although most of cases were non-Hodgkin’s lymphoma and multiple myeloma. Second, although history of previous viral infections was excluded, there were still viral infections whose compatible ICD-9-CM code could not be found, including EBV and HHV-8. However, since Taiwan is known an endemic area of EBV infection and most adults are EBV-infected
, and HHV-8 related primary effusion lymphoma is very rare, it is likely that the influence of un-coded viral infections in patients and control group is similar. Third, the possibility of some herpes zoster infection being missed or inaccurately diagnosed exists. However, it is likely that the possibility did not differ between cases and controls, and herpes zoster has unique clinical presentations. Forth, analyzing the combined outcomes of all malignancies and lymphoid malignancies in both groups may be a limitation because there are many distinct entities with different etiologies of lymphoid malignancies. As such what we measured is an average association between herpes zoster and all lymphoid malignancies. Lastly, although several adjusted analyses had been performed to exclude the influence of medical comorbidities listed in CCI, there may be unmeasured confounding effects by other medical illness that have not been captured. The possibility that a few HIV-infected people being missed in the claims database is low because HIV-infected patients before diagnosis of any malignancies were excluded; HIV exam is an essential part when lymphoid malignancies are diagnosed; no patients were found to have HIV infection during follow-up; and HIV-infected patients have a mark in their own insurance card and they will not get the insurance coverage for HIV management if no HIV ICD-9-CM code was claimed.