Immune Restoration Disease After Antiretroviral Therapy

Martyn A. French; Patricia Price; Shelley F. Stone


AIDS. 2004;18(12) 

In This Article

Infectious Immune Restoration Disease

The observation of atypical presentations of Mycobacterium avium complex (MAC) disease in patients treated with zidovudine monotherapy was the first indication that restoring pathogen-specific immune responses can cause immunopathology.[5] MAC disease in this situation was localized rather than disseminated and was characterized by severe fever, painful lesions and a granulomatous inflammatory response or suppuration. Most importantly, it was associated with restoration of cutaneous delayed-type hypersensitivity (DTH) responses to mycobacterial antigens, whereas MAC disease before antiretroviral therapy had invariably been associated with anergy. Furthermore, subsequent loss of DTH responses to mycobacterial antigens was associated with the occurrence of disseminated MAC infection.[16]

MAC IRD in patients responding to HAART presents with fever, lymphadenitis, pulmonary infiltrates or inflammatory masses.[6,17,18,19,20,21,22,23,24,25,26] Pyomyositis and cutaneous abscesses have also been described.[27] MAC IRD usually occurs during the first three months of therapy in patients with pre-therapy CD4 T-cell counts < 100 × 106 cells/l. Lymphadenitis is often painful and may suppurate. Inflammatory masses are usually endobronchial but may affect the abdominal cavity. Histological examination of lymph nodes or inflammatory masses often reveals granulomatous inflammation. MAC is usually cultured from lesions but not from the blood. Cutaneous DTH responses to mycobacterial antigens are a characteristic finding.[6,19,25]

Restoration of cellular immune responses to Mycobacterium tuberculosis (MTB) antigens may also cause IRD in patients responding to HAART.[28,29,30,31,32,33,34,35,36] MTB IRD is associated with the restoration of a cutaneous DTH response to tuberculin.[28,29,34] It presents within the first 2 months of HAART, usually in the first 2-3 weeks. Common presenting features are severe fever, intra-thoracic and cervical lymphadenopathy and pulmonary infiltrates. Extra-thoracic disease is less common and includes focal cerebritis,[6,30] pleural effusions, hepatosplenomegaly and ascites.[31] Pulmonary disease may be associated with hypercalcaemia.[36]

Bacille Calmette-Guerin (BCG) lymphadenitis in a child, which was associated with a DTH response to tuberculin,[37] and borderline tuberculoid leprosy[38] have also been described in patients responding to HAART.

Several lines of evidence indicate that mycobacterial IRD results from a DTH response to mycobacterial antigens. Granulomatous inflammation and tissue necrosis are typical of DTH responses.[39] Foudraine et al.[40] showed that MAC disease after HAART was associated with increased lymphoproliferation responses to mycobacterial antigens. The occurrence of MAC IRD and, to our knowledge, the lack of virus-associated IRD after zidovudine monotherapy supports the involvement of DTH because zidovudine monotherapy is sufficient to augment DTH responses.[41] Finally, a case study showed a correlation between MAC IRD and DTH responses to tuberculin in a patient who never had CD4 T-cell deficiency.[25]

The basic elements of managing mycobacterial IRD are anti-mycobacterial and anti-inflammatory therapies. Corticosteroids are usually used but pentoxyphylline may be effective for MTB IRD.[28]

Soon after the introduction of HAART, it was observed that some patients presented with an initial or recurrent episode of cryptococcal meningitis during the first few weeks of therapy.[42] It occurred in patients with CD4 T-cell counts < 50 × 106 cells/l who had an immunological and virological response to HAART and was characterized by a more prominent inflammatory cell reaction in the CSF than in patients not receiving HAART. Cryptococci were usually cultured from CSF and/or blood. Subsequently, Jenny-Avital et al. reported that 50% of patients with cryptococcal infection who responded to HAART developed cryptococcal disease despite the use of fluconazole therapy.[43] However, cryptococcal disease in this series of patients presented 2 to 11 months after commencing HAART and was characterized by failure to culture cryptococci even though organisms were seen in lesions or cryptococcal antigen was detected in body fluids. Reports on other individuals or small series of patients have shown similar findings and a pattern of clinical presentation of late cryptococcal IRD is emerging.

The most common presentation of late cryptococcal IRD is lymphadenitis, particularly mediastinal lymphadenitis.[43,44,45,46] It is reported to present up to 15 months after commencing HAART.[44] Histological examination of nodes may reveal granulomatous inflammation, necrosis or suppuration. Granulomatous inflammation may be associated with hypercalcaemia.[43] Cryptococcal IRD of the central nervous system (CNS) has also been described. Aseptic meningitis is associated with high intracranial pressure[43,47] and localized inflammatory lesions may occur in the spinal cord[48] or brain.[49]

There are no known data showing that cryptococcal disease in patients responding to HAART is associated with increased immune responses to cryptoccocal antigens. However, an interpretation of the available clinical data suggests that effective HAART may result in two types of inflammatory reaction to cryptoccoci. Firstly, an acute inflammatory reaction to subclinical meningeal infection during the first few weeks of therapy and secondly, a chronic inflammatory reaction in lymph nodes or the CNS, possibly initiated by a T-cell response to the antigens of non-viable cryptoccoci.

The optimal management of cryptococcal IRD has not been defined. Anti-inflammatory therapy has been used[44] but many cases resolve spontaneously.

Pneumocystis carinii pneumonitis that has improved on anti-Pneumocystis therapy may relapse following the introduction of HAART.[50,51,52] Commencement of HAART within 18 days of commencing anti-Pneumocystis therapy and/or cessation of corticosteroid therapy appear to be risk factors, but only a limited amount of data are available.

Hepatotoxicity is an adverse effect of HAART in up to 18% of patients. This is a direct effect of antiretroviral drugs, particularly nevirapine and high-dose ritonavir, in some patients but hepatotoxicity occurs most often in patients with hepatitis B virus (HBV) or hepatitis C virus (HCV) co-infection (for example see[53,54,55]). The pathogenesis of hepatotoxicity in HIV/HCV or HIV/HBV co-infected patients has not been defined, but early clinical studies suggested IRD might be a cause.[56,57]

Liver biopsies in HIV/HCV co-infected patients with HAART-associated hepatotoxicity show changes of viral hepatitis,[54,57] suggesting that there has been an immune response against the HCV. However, the association of hepatotoxicity with increased blood CD4 T-cell counts is controversial[54,55,58] and there are no known data on HCV-specific T-cell responses. Studies of pathogenic mechanisms may require the use of liver biopsy samples or serological markers to demonstrate the role of HCV-specific immune responses because the cellular immune response in HCV-associated hepatitis is compartmentalized to the liver.[59] For example, HAART-associated hepatotoxicity is associated with increased plasma levels of HCV antibodies[57,60,61] and dipeptidyl dipeptidase IV [DPP IV, soluble (s) CD26] activity.[61] sCD26 (DPP IV) is an aminopeptidase which down-regulates CXCR3,[62] a major homing molecule expressed by Th1 cells in the HCV-infected liver.[63]

Although HAART-associated hepatotoxicity in HIV/HCV co-infected patients may result in fulminant hepatitis and sometimes cirrhosis,[64,65] this is very uncommon.[54] The outcome for most patients appears to be similar to patients who do not experience hepatotoxicity.[66,67] However, the long- term effect of hepatotoxicity on liver fibrosis is currently unknown and it has been argued that HCV infection should be treated before commencing HAART to reduce the probability of hepatotoxicity.[68]

HAART-associated hepatotoxicity in HIV/HBV co-infected patients is associated with a decreased plasma HIV RNA level and increased CD4 T-cell count, and liver biopsies show changes of viral hepatitis.[56,69,70,71,72,73] In some patients, hepatotoxicity was associated with clearance of HBeAg and HBV DNA and increased anti-HBc and anti-HBe[56,70,73] whereas in others it was associated with increased plasma levels of HBV DNA, reappearance of HBsAg, loss of anti-HBs and appearance of anti-HBcIgM.[69,71,72] An explanation for the dichotomous immunological and virological effects of HAART in HIV/HBV co-infected patients with hepatotoxicity is not available. Clearly, further studies are needed to elucidate pathogenic mechanisms.

Eye disease is the most common presentation of CMV IRD. It presents as retinitis in the first 3 months of HAART or as uveitis, usually much later than retinitis. CMV retinitis IRD may be the first indication of CMV infection,[74,75] or present as a relapse of established CMV retinitis that is in remission on anti-CMV therapy[6,76] and subsequently resolves without relapse when anti-CMV therapy is ceased.[77] It occurs in patients with CD4 T-cell counts < 100 × 106 cells/l who experience an increased CD4 T-cell count in response to HAART. A positive plasma CMV polymerase chain reaction was a strong predictor of CMV retinitis in a prospective study of patients commencing protease inhibitor therapy.[75]

Immune recovery uveitis (IRU) is an inflammatory disorder that affects eyes with inactive CMV retinitis in patients who have experienced an increased CD4 T-cell count on HAART.[78,79,80,81] It presents with floaters and/or impaired visual acuity and can result in permanent visual impairment. The inflammatory changes may include vitritis, papillitis, cystoid macular oedema and epiretinal membranes. Fibrovascular membranes resulting from the inflammation may be misdiagnosed as recurrent CMV retinitis.[82] Karavellas et al.[81] reported that IRU presents 2-84 (median 20) weeks after an increase in the CD4 T-cell count. Eyes with an average CMV retinitis area of > 30% had a 4.5-fold increased risk of developing IRU. The incidence was estimated to be 0.83/person-year. However, this is much greater than the rate of 0.109/person-year reported in another study.[80] IRU can cause severe inflammatory eye disease and, therefore, patients on HAART with a previous history of CMV retinitis who develop visual symptoms must be assessed by an ophthalmologist. Topical or systemic corticosteroid therapy is usually effective.

First presentations or relapses of CMV retinitis after commencing HAART appear to represent the effects of an immune response against retinal CMV infection that eventually controls the infection. However, Hsieh et al.[83] showed that CMV-specific CD4 T-cell responses were lower in patients who developed CMV retinitis after commencing HAART compared with patients without retinitis. We and others[84,85] have shown that CD4 T-cell responses to CMV antigens remain low in all patients with nadir CD4 T-cell counts of < 50 × 106 cells/l during the first year of HAART. However, we have also shown that CMV retinitis after HAART is associated with increased plasma levels of IgG anti-CMV antibody and soluble (s) CD30[86] and with increased bioavailable interleukin (IL)-6.[87] Our interpretation of these findings is that the immune response restored against CMV antigens in patients with CMV retinitis IRD is immunopathological and characterized by a Th2 bias and/or a CD8 T-cell response because CD30 is expressed by Th2 cells and has a regulatory function in CD8 cytolytic T cells.[88] With regard to this, Hsieh et al.[83] showed that CMV-specific CD8 T-cell responses in patients with CMV retinitis after HAART are similar to patients without retinitis, indicating that there is a relative predominance of CD8 over CD4 CMV-specific T-cell responses in patients who develop retinitis.

Examination of epiretinal membranes from eyes with IRU demonstrated the presence of chronic inflammatory cells consisting predominantly of T cells.[81] Preliminary evidence suggests that the inflammation represents a CMV-specific CD8 T-cell response. Thus, cloning of T cells from the vitreous fluid of a single patient demonstrated a predominance of CMV-specific CD8 T cells[89] and the severity of IRU is not related to the magnitude of the increase in blood CD4 T cells.[81]

Rates of varicella zoster virus (VZV) disease are increased in patients responding to HAART. Dermatomal VZV disease occurs in 6-8% of patients at a rate of 6.2-9.0 episodes per 100 patient years.[6,90,91,92] The rate has been estimated to be five times higher than expected.[90] The majority of cases present during the first 4 months of HAART but some cases present much later.

Studies by two Spanish groups have identified a greater increase in CD8 T-cell counts as a risk factor for dermatomal VZV disease after HAART,[91,92] suggesting that CD8 T cells may be mediating an immune response against VZV antigens. A case study of transverse myelitis resulting from VZV IRD demonstrated that NK cells as well as CD8 T cells may be mediating a response to VZV antigens.[93] Of note, the immune response in that case was compartmentalized to the CNS, indicating that IRD may occur in the absence of immune reconstitution in the blood.

VZV IRD responds to acyclovir therapy[6,91,92] suggesting that the immune response is against active VZV infection. However, VZV stromal keratitis has also been reported as an IRD in a patient receiving prophylactic acyclovir therapy.[94]

Data from an observational study suggests that mucocutaneous herpes simplex virus (HSV) disease may occur more frequently and be more severe in patients responding to HAART.[6] In addition, Fox et al.[95] reported chronic penile ulceration related to HSV infection in black African men. Of note, all three affected patients carried HLA-B72, -Cw0202 and -DRB4, suggesting that there is a genetic susceptibility to this condition. Presumptive HSV infection has been associated with encephalomyelitis after HAART.[6]

HAART is the only effective therapy for progressive multifocal leukoencephalopathy (PML) in HIV patients,[96] probably because it augments CD4 T-cell responses to JC virus (JCV) antigens.[97] However, in some patients who commence HAART, PML may become worse or present for the first time.[96,98,99,100,101,102,103] Cinque et al.[96] reported that 18% of patients with PML experienced an exacerbation within 9 weeks of commencing HAART associated with a reduction of the plasma HIV RNA level and an increased CD4 T-cell count. Antiretroviral-naive patients responding to HAART had the worst outcome with 50% experiencing progression of PML lesions. Histopathology examination of lesions demonstrates more inflammation than in lesions from patients not receiving HAART. In particular, perivascular infiltrations of lymphocytes, monocytes and plasma cells have been observed.[99,101,102,103] The lymphocytes are predominantly CD8 T cells,[99,101] suggesting that CD8 T cells may mediate an immunopathological response to JCV antigens. Supporting evidence has been provided by a case study of a patient who developed PML after responding to HAART without an increase in the blood CD4 T-cell count.[103]

Evidence-based guidelines for the prevention and management of JCV IRD are not available. However, there is preliminary evidence that modification of the antiretroviral regimen might have a beneficial effect. In the series of patients reported by Cinque et al.,[96] a better outcome was observed in patients who interrupted therapy or used a drug regimen that resulted in slower immune reconstitution. Corticosteroid therapy has been used to treat JCV IRD but may be ineffective.[101,102]

There are several case reports of IRD associated with uncommon opportunistic pathogens. These are listed in Table 1 .

Other diseases associated with an infection by an opportunistic viral pathogen are reported to present atypically or more frequently in patients responding to HAART. The evidence that they are IRD is not as strong as for the conditions discussed so far, but they are discussed to promote further study of their pathogenesis because an immunopathological response to the virus might be involved.

Two publications documenting an increase in the prevalence of oral warts in the context of a decrease in Candida stomatitis, hairy leukoplakia and Kaposi's sarcoma of the mouth in patients on HAART[111,112] raise the possibility of human papillomavirus IRD. Non-Hodgkin's lymphoma (NHL) after commencing HAART[113] might be related to an immunopathological response to Epstein Barr virus (EBV) infection, as T-cell responses to some EBV antigens induce IL-10 production[114] and increased IL-10 production is associated with HIV-associated NHL[115]. Finally, reports that Kaposi's sarcoma[116] or multicentric Castleman's disease[117] develop after commencing HAART suggest a possible immunopathological response to human herpes virus-8.

Collation of the clinical, immunological and immunogenetic data presented over the last 12 years has identified risk factors for infectious IRD that could be used in a risk assessment for patients commencing HAART (Fig. 1). Firstly, an active or subclinical infection by opportunistic pathogens,[16] or the antigens of non-viable micro-organisms (eg. cryptococci and CMV) are all possible targets for an immunopathological response. This may explain why prophylactic anti-MAC therapy may not prevent MAC IRD.[118] Secondly, a CD4 T-cell count below 50 × 106 cells/l is a major risk factor for IRD.[6] It may be a marker of a high pathogen load and/or an increased susceptibility to immune dysregulation during immune reconstitution. Finally, there are disease susceptibility genes for some IRD. The first indication of this was that some patients with CMV retinitis presenting as IRD also experienced neurological disease associated with definitive or presumptive infection by other herpes viruses.[6] This suggested a genetic susceptibility to an immunopathological response against herpes viruses. Supportive evidence has been provided by our studies showing that HLA-B44 and the major histocompatibility complex (MHC) ancestral haplotype HLA-A2, -B44, -DR4,[119] and allele 1 at a single nucleotide polymorphism (SNP) in the 3'UTR of the IL12B gene encoding IL-12 p40[120] are associated with herpes virus IRD. The significance of these findings was strengthened by the observation that alleles of other cytokines genes (TNFA-308*2 and IL6-174*G) were not associated with herpesvirus IRD but were associated with mycobacterial IRD.

Risk factors for infectious immune restoration disease (IRD).

If advances are to be made in the diagnosis and management of infectious IRD, it is essential that diagnostic criteria are formulated and tested in clinical practice. Proposed diagnostic criteria are outlined in Table 2 . These are based on the findings of a retrospective study[6] and the many clinical studies referred to in this review. A diagnosis of IRD would require both major criteria or criterion A and two minor criteria.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: