Influenza in Immunosuppressed Populations

Abstract

Patients that are immunosuppressed might be at risk of serious influenza-associated complications. As a result, multiple guidelines recommend influenza vaccination for patients infected with HIV, who have received solid-organ transplants, who have received haemopoietic stem-cell transplants, and patients on haemodialysis. However, immunosuppression might also limit vaccine responses. To better inform policy, we reviewed the published work relevant to incidence, outcomes, and prevention of influenza infection in these patients, and in patients being treated chemotherapy and with systemic corticosteroids. Available data suggest that most immunosuppressed populations are indeed at higher risk of influenza-associated complications, have a general trend toward impaired humoral vaccine responses (although these data are mixed), and can be safely vaccinated—although longitudinal data are largely lacking. Randomised clinical trial data were limited to one study of HIV-infected patients with high vaccine efficacy. Better trial data would inform vaccination recommendations on the basis of efficacy and cost in these at-risk populations.

Introduction

The influenza virus is among the most common human respiratory viruses. In the USA, for example, influenza can result in the admission of over 225 000 patients to hospital¹ and 36 000 deaths every year.² High rates of influenza infection and complication rates are suggested to occur among people with impaired immune defences. Immunosuppression can result from many different biological mechanisms, ranging from rare congenital immunodeficiencies to more common causes such as immunosuppressive drugs or HIV infection.

Despite prevailing concerns about the risk of influenza infection in patients with immunosuppression, the actual rate and effect of infection in this population are not well characterised. Whether people that are immunocompromised would be even more susceptible to new viruses, such as the recent swine-origin influenza A H1N1 strain,³ without the benefit of previous exposure, is an area for investigation. Moreover, although influenza vaccination is widely recommended for people that are immunosuppressed, the same immune dysfunction that can increase the risk and consequences of influenza infection might also compromise vaccine responses and effectiveness. These distinctions are particularly important in settings of vaccine shortages or in response to a new pandemic, where selective and prioritised allocation of vaccine is necessary. To better inform influenza vaccination policies for immunocompromised populations, we reviewed the published work relevant to influenza infection and vaccination in patients with HIV infection, recipients of solid-organ transplants (SOTs), recipients of haemopoietic stem-cell transplants (HSCTs), patients receiving chemotherapy, patients on chronic haemodialysis, and patients receiving corticosteroids. These conditions affect about 3 million people in the USA (about 1% of the total US population), and the worldwide HIV epidemic affects about 33 million people (table 1).

Table 1. Estimated prevalence of selected immunosuppressive conditions.

	Condition	USA	Worldwide/Other
Chronic immunosuppressive conditions	HIV/AIDS4	1,300,000a, *	33,200,000a
	Solid Organ Transplant5	153,245c	Data not found
	Hemodialysis6	314,162b	Data not found
Transient immunosuppressive conditions	Hematopoetic Stem Cell Transplant (annual procedures)7	18,720b, *	45,000b
	Cancer (new cases)**8	1,479,350e	Data not found
Either chronic or transient	Corticosteroid use9	Data not found	409,000 in U.K.d

Superscript letters indicate data for years as follows

^a2007

^b2005

^c2004

^d1996

^eestimate for 2009

^*Data for all of North America

^**Data for hematologic and invasive solid organ cancers only; excludes in-situ malignancie

The primary purpose of this Review is to identify and summarise the incidence and mortality rates of influenza infection among adults who are immunocompromised, as well as the risks of vaccination, the ability of vaccine to elicit appropriate immune responses, and the clinical effectiveness of vaccination in these populations. We also summarise guidelines on influenza vaccination of these populations, limitations of the available data, and alternative approaches to influenza control.

HIV/AIDS

An estimated 1·3 million (480 000 to 1·9 million) people are living with HIV/AIDS in North America. Worldwide, the epidemic affects an estimated 33·2 million (30·6 million to 36·1 million) people.⁴ Influenza is a common cause of respiratory illness in adults with HIV.¹⁰ Because both antibodies and T cells play a crucial part in viral immune responses, patients infected with HIV might be expected have severe and prolonged influenza infections, as reported in several early case series and anecdotal reports.^11-14

Two large studies done before the widespread use of highly active antiretroviral therapy (HAART) confirmed high numbers of hospital admissions and mortality from influenza in patients with HIV/AIDS (table 2).^15,16 A study analysing data from an administrative database¹⁷ showed that cardiopulmonary hospitalisation rates of patients with HIV/AIDS decreased by 53% during the HAART era. However, the reported hospitalisation rate still exceeded that of the general US population, and was similar to that of other high-risk populations for whom the US Centers for Disease Control (CDC) recommends influenza vaccination.¹⁸ Although HAART seems to reduce the number of patients with HIV/AIDS treated for influenza in hospital, much of the world’s population living with HIV/AIDS are in resource-limited areas such as sub-Saharan Africa, where antiretroviral drugs are available to less than 10% of those in need of treatment.¹⁹

Table 2. Incidence of influenza in general and immunosuppressed populations.

Author	Data source	Population	n	Outcome	# Outcomes/1000 person-years in disease (+) population (95% CI)	# Outcomes /1000 person-years in control (95% CI)
Thomp son2	National Center for Health Statistics Database	General population	nr	Influenza-associated deathsb	n/a	All ages: 0.03 (nr) ≥65 y/o: 0.22 (nr)
Thomp son1	National Hospital Discharge Survey	General population	nr	Influenza-associated hospital admissionsc	n/a	All ages: 0.52 (nr) >65 y/o: 2.81 (nr)
Neuzil15	Tennessee Medicaid Database, women 15-64 y/o for HIV and for controls.	HIV, Pre-HAART	1,725,712 person-years of data	Combined, excess hospital admissions and deaths	1973-1993: 33.4 (nr)	0.4 (nr)
Lin16	National Center for Health Statistics Database, ≥13 y/o. for AIDS, 25-54 y/o for controls	AIDS, Pre-HAART	149,256 person-years of data	Excess deaths from pneumonia or influenza during influenza seasonsa	1991-1992: 1.26 (1.07-1.45) 1992-1993: 1.47 (1.28-1.65) 1993-1994: 0.94 (0.80-1.08)	0.01 (0.009-0.01) 0.009 (0.009-0.01) 0.009 (0.008-0.009)
Neuzil17	Tennessee Medicaid Database. HIV+, 15-50 y/o.	HIV, Post-HAART	7,368 person-years of data	Excess hospital admissions Excess deaths	1995-1996, pre-HAART: 48 (16-81) 1996-1999, post-HAART: 5 (-0.5-11) 0.5 (-4.8-5.7)	n/a
Vilchez62	University of Pittsburgh Transplant Program Database.	Organ Transplant	3,569 organ transplants over 10 years	Hospital admissions	Lung Tx: 4.18 - 41.8 (nr)d Liver Tx: 0.28 - 2.8 (nr)d Kidney Tx: 0.43 - 4.3 (nr)d	n/a

n/a : not applicable

nr: not reported or data not collected

^aExcess deaths calculated as difference between deaths during influenza season and prior to influenza season

^bData reported are for underlying pneumonia and influenza deaths. Data for all-cause deaths showed rates of 0.20 for all ages, and 1.32 for ≥65 y/o.

^cData reported are for influenza or pneumonia listed under any diagnosis. Data for influenza as primary diagnosis showed 0.37 endpoints/1000 person-years for all ages.

^dRanges provided because complete person-years data not reported.

Yearly influenza vaccination for adults infected with HIV is recommended by the CDC and the Infectious Disease Society of America,^18,20,21 although this recommendation has not received universal support. Initial concerns that immunisation enhances HIV replication by vaccine-induced CD4+ T cell activation²² have been supported by some,^23-30 but not all,^31-38 studies. Moreover, rises in plasma HIV RNA that occur do so only transiently and then return to baseline over several weeks, typically without associated changes in CD4+ T cell counts, time to AIDS, or death over 8 years of follow-up in one retrospective analysis.³⁹ No prospective studies have looked at the long-term effect of influenza vaccination on disease progression or the effect of annual vaccination over multiple years.

Although influenza vaccination of patients with HIV/AIDS seems to be safe, the T-cell dysfunction caused by HIV might be expected to diminish the immune response to influenza vaccination. Indeed, multiple studies have shown lower antibody responses in patients infected with HIV compared with the uninfected people in the control group,^{23,35,36,40-44} although responses in one study were similar.³³ The prediction that patients with low CD4+ counts or with a diagnosis of AIDS would have the most limited antibody responses to immunisation is supported by several studies,^32,41,43,45 but not all reports (figure).^35,36,44 Most of these studies are limited by small sample sizes, particularly for patients with very low CD4+ counts; the five studies reporting responses in patients with CD4+ counts of less than 200 CD4+ T cells per μL included only 55 patients. Two recent HAART-era studies showed vaccine response to be related to HIV RNA levels rather than CD4+ counts.^55,56

Figure. Percentage of immunosuppressed patients and healthy controls with protective postvaccination influenza titres.

Data are only presented for proportions of patients with H3N2 serotype of influenza A, which is typically associated with higher risk of admission to hospital and death than the H1N1 serotype or influenza B.¹,² Protective haemagglutination inhibition (HI) titre for studies was 1:40 or greater unless otherwise noted. Studies without healthy control groups are not presented. No studies provided data on haemopoietic stem-cell transplant recipients relevant to these figures. Also excluded are data for corticosteroid users, because of inability to extract relevant data for figures. Results in patients with HIV infection (A), stratified by CD4+ T cell counts, and healthy controls. Fowke and colleagues³² not shown, because CD4 strata were not readily integrated with more conventional CD4 strata above. Brydak and colleagues⁴⁴ did not include healthy control group, but data were provided for CD4 strata, and therefore included. Results in solid-organ transplant recipients and healthy controls (B): Hayney and colleagues⁴⁶ data extracted from published figures. Dengler and colleagues⁴⁷ reported percentage with either HI titer 1:40 or greater or with four-fold or greater rise in HI titre postvaccination. Birdwell and colleagues⁴⁸ reported HI titer of 1:32 or greater. Results in patients treated with chemotherapy for malignancies and healthy controls (C): Feery and colleagues⁴⁹ and Gribabis and colleagues⁵⁰ data extracted from published figures. Stiver and colleagues⁵¹ reported HI titre of 1:32 or greater. Results in patients on hemodialysis and healthy controls (D): Beyer and colleagues⁵² reported HI titre of 1:100 or greater. Rautenburg and colleagues⁵³ data extracted from published figures and reported IgG titre of 1:2 or greater. Antonen and colleagues⁵⁴ data extracted from published figures.

Clinical protection from disease (immunity), rather than antibody response, is the ideal outcome for vaccine studies. An outbreak investigation at a residential facility for people with AIDS, found fewer and less severe influenza cases in people previously vaccinated than in those not vaccinated.⁵⁷ Two prospective, but nonrandomised, studies identified substantially fewer episodes of laboratory-confirmed influenza in vaccinated patients with HIV (6·1%) than in unvaccinated patients (21·2%),⁵⁶ and a lower frequency of influenza in patients who were simultaneously vaccinated for influenza and pneumococcus than in those who declined vaccination.⁵⁸ The only randomised, double-blind, placebo-controlled study showed a 20% absolute reduction in risk of respiratory symptoms (from 49% to 29%, p=0·04) and 100% protection (95% CI 73-100%) against laboratory-confirmed symptomatic influenza (p<0·001) among patients infected with HIV receiving influenza vaccine compared with those receiving placebo.³³ Two systematic reviews identifying the same four clinical outcome studies cited above, conclude that influenza vaccination of adults infected with HIV might be effective despite variable antibody responses.^59,60 Both reviews highlight the limited data available and the need for more and larger randomised trials (only one has been done to date) to confirm how effective and safe vaccination is in this large population of patients, particularly among those with very low CD4+ counts.

Solid-organ transplantation

Patients receiving SOTs are treated with immunosuppressive drugs that alter multiple immune mechanisms.⁶¹ Therefore, those patients might have more severe influenza infections and compromised responses to vaccination. As of the end of 2004, 153 245 people in the USA were living with a functioning transplanted solid organ.⁵ Over 27 000 transplantations are done every year in the USA, nearly 90% of which are kidney, liver, and heart procedures.⁵

A retrospective review of patients that received organ transplants admitted to hospital with influenza between 1990 and 2000 reported high numbers of patients who received lung transplants compared with patients that received liver and renal transplants (table 2).⁶² These increased rates might be because of either increased surveillance of recipients of lung transplants, decreased secretory antibodies in the allograft lung, epithelial injury, or impaired lymphatic drainage of the allograft lung—the primary site of influenza infections. Other studies in recipients of lung transplants with acute viral respiratory symptoms have identified influenza in 3%,⁶³ 16%,⁶⁴ and 48%⁶⁵ of viral illnesses during times when influenza is circulating in the community.

Prevention of influenza infection in recipients of SOTs is a crucial issue. Influenza infection has been reported to cause acute kidney allograft rejection,^66-68 and both acute^62,69 and chronic lung allograft rejection.^69-71 Despite these potentially devastating consequences of infection in recipients of SOTs, influenza vaccination in this population could stimulate a T-cell response, thus leading to organ rejection. Multiple studies of kidney^72-80 and heart^47,81-85 transplantation have not linked allograft rejection to influenza vaccination. However, all studies were underpowered to derive firm conclusions on safety.

The effectiveness of influenza vaccination of recipients SOTs is also widely debated. Despite some evidence for comparable influenza vaccine responses in recipients of kidney^72,80,86,87 and liver⁸⁸ transplants and control individuals, most published data suggest that antibody responses are diminished in recipients of SOTs, whether the transplanted organ is a kidney,^{48,73,75,76,78,79,89,90} liver,^91-93 heart,^47,83,92,94 or lung^46,95 (figure). The particular immunosuppressive regimen might contribute to the size of any impairment. Azathioprine, for example, might have less effect on serological responses than mycophenolate mofetil^46,76,86 or ciclosporin.^77,90 Indeed, mycophenolate mofetil use has been associated with lower serological response rates compared with other immunosuppressive drugs.^55,72 Data that recipients of SOTs treated with sirolimus and those treated with a calcineurin inhibitor (tacrolimus or ciclosporin) had similar serological responses to each other⁹⁶ might have been underpowered (n=32 patients) to detect substantial differences. Despite their lower overall response rates to influenza vaccination, many recipients of SOTs do mount an appropriate immune response. Evidence of the clinical effectiveness of immunisation is lacking for recipients of SOTs.

The American Society of Transplantation recommends influenza vaccination every year for all recipients of SOTs, beginning about 6 months after transplantation.⁹⁷ 6 months was chosen because immunosuppressive protocols generally have been tapered substantially by that time and a recent study showed lower antibody titres in recipients of kidney transplants vaccinated within 6 months of transplantation.⁴⁸ Despite these recommendations, among 1800 recipients of SOTs transplanted between 1995-2005, nearly half were unvaccinated.⁹⁸

Bone-marrow transplantation

An estimated 35 000 people receive autologous (reimplanted from the same person) and 20 000 allogeneic (from a genetically similar donor) HSCTs every year worldwide.⁷ Preparation for stem-cell infusion generally requires high doses of chemotherapy, radiation, or both to prevent allograft rejection from residual host immune cells, destroy any residual malignant cells in patients receiving HSCTs for malignancies, and provide marrow space for donor stem-cell engraftment. These intensive pretransplantation regimens leave patients acutely and profoundly immunocompromised for up to several months after transplantation. Additional complications, such as delayed engraftment and graft-versus-host disease (GVHD), can predispose these patients to more prolonged immune dysfunction.

The frequency of influenza among recipients of HSCTs is quite low (range of 0·2-2·8%; table 3).^100-105 However, among patients with respiratory symptoms during a local epidemic, influenza frequency is as high as 23-29%.^104,105 Individual reports have suggested an increased frequency of influenza illness among recipients of allogeneic HSCTs compared with recipients of autologous HSCTs.^101,102 However, the largest analysis so far—a 12-year, single-centre, retrospective analysis of 4797 patients—found no association between donor type and risk of influenza within the first 120 days after transplantation.¹⁰³ In this same analysis, the presence of GVHD was not associated with increased risk of influenza infection. Additionally, influenza might have devastating consequences in recipients of HSCTs. The largest report (37 European centres, three influenza seasons, over 1900 patients) cited a case fatality rate of 23% (table 3).¹⁰¹

Table 3. Influenza frequency and case fatality in HSCT and cancer patients.

Author	Population and Setting	Influenza Years	# of patients under observation	Frequency of influenza†	Case fatality
Couch99	HSCT, hospitalized, single center	1992-1995	nr	nr	5/20 (25%)
Couch99	Leukemia, hospitalized, single center	1992-1995	nr	nr	9/27 (33%)
Ljungman100	HSCT, single center.*	1989-1996	545	15 (2.8%)	2/15 (13%)
Ljungman101	Allogeneic HSCT , 37 European Centers* Autologous HSCT, 37 European Centers* Allogeneic HSCT, 37 European Centers* Autologous HSCT, 37 European Centers*	1997-1998 1997-2000	819 1154 >819 >1154	14 (1.7%) 2 (0.2%) nr nr	4/14 (29%) 0/2 (0%) 7/30 (23%) 2/9 (22%)
Hassan102	Allogeneic HSCT, single center* Autologous HSCT, single center*	1996-2001	230 396	5 (2.2%) 0 (0%)	1/5 (20%)
Nichols103	HSCT, 120 days within transplantation date only, single center*	1989-2002	4797	62 (1.3%)	6/62 (10%)
Machado104	HSCT, respiratory symptoms present, single center*	2001-2002	179	41/179 (23%)	0/41 (0%)a
Whimbey105	HSCT, hospitalized, local influenza epidemic present, respiratory symptoms present, single center	1991-1992	28	8/28 (29%)	1/8 (13%)
Yousuf106	Leukemia, hospitalized, local influenza epidemic present, respiratory symptoms present, single center	1993-1994	45	15/45 (33%)	4/15 (27%)
Schepetiuk107	Nosocomial outbreak in oncology ward, single center	1997	19	nr	2/19 (11%)
Elting108	Leukemia, local influenza epidemic present, respiratory symptoms present, single center.	1991-1992	37	4/37 (11%)	1/4 (25%)

^†Frequencies are those reported during the total observation period (column 3—influenza years)

^*outpatient/inpatient status not specified.

^aonly reported mortality from pneumonia, not all-cause mortality

nr = not reported

Very few controlled studies of influenza vaccination in recipients of HSCTs have been done, making vaccine responses and efficacy difficult to establish. One study found overall poor serological responses to influenza vaccine: from 0% in nine recipients of allogeneic transplants to 32% in 41 recipients of autologous transplants.¹⁰⁹ The difference in serological response rates by donor type was not statistically significant in this small study. Another study similarly showed blunted responses to influenza vaccine for recipients of HSCTs compared with healthy controls (29-34% vs. 46-62%, respectively),¹¹⁰ but again, small sample sizes limited statistical power. Addition of granulocyte-macrophage colony-stimulating factor to the vaccination regimen did not improve serological responses.

In addition to regimen, appropriate timing might be important for the immunogenicity of influenza vaccination after giving HSCTs. One series showed that vaccination within the first 6 months resulted in a complete lack of serological response in all ten patients.¹¹¹ For vaccination within 7-24 months after receipt of HSCT, the response rate was 13% and reached 64-71% for vaccination 2 years after the procedure.

We found only one retrospective report of the clinical effectiveness of influenza vaccination in recipients of HSCTs.¹¹² Among patients with respiratory symptoms who received a nasal wash or aspirate, 134 people were unvaccinated (because of being less than 6 months from transplantation), and 25 of them (19%) developed influenza. Among the 43 patients who were beyond 6 months from transplantation, influenza was diagnosed in 50% of the 24 unvaccinated patients (because of non-compliance with the programme), but only 11% of the 19 vaccinated patients. No prospective or randomised studies have been done to confirm or quantify vaccine efficacy in this population at high risk for adverse outcomes with influenza.

The CDC, Infectious Disease Society of America, and American Society of Blood and Marrow Transplantation have jointly published guidelines for influenza vaccination of recipients of HSCTs,¹¹³ as has the European Group for Blood and Marrow Transplantation.¹¹⁴ Both sets of guidelines support influenza vaccination beginning around 6 months after transplantation, but they differ in the recommended periodicity of immunisation (lifelong in the US guidelines vs. individually assessed duration in the European guidelines). Unfortunately, no data exist to help resolve this difference, since most studies have only investigated the effect of a single dose given in one influenza season. No study has looked at the long-term risks or benefits of annual influenza vaccination in this immunosuppressed population.

Malignancies and chemotherapy

Some 1·4 million patients will be diagnosed with new haematological and invasive solid-organ cancers in the USA in 2009,⁸ many of whom will require treatment with chemotherapy. Similar to preparation for HSCT, chemotherapy can produce acute and profound immunosuppression, although the degree differs depending on the chemotherapy drugs, doses, and total duration of therapy. An estimated number of patients receiving chemotherapy every year is not available.

Information about influenza frequency among adults receiving chemotherapy is quite limited. Among patients with cancer admitted to hospital with respiratory symptoms during times of a circulating influenza epidemic, 21-33% might test positive for influenza.^108,106 Similar to results in recipients of HSCTs, influenza mortality is very high among patients with cancer. Four different studies reported case fatalities of 11%, 25%, 27%, and 33% (table 3).^99,107,108 The available data are too limited to distinguish frequency and mortality differences among patients with different malignancies and different chemotherapeutic regimens.

As in the HSCT population, humoral responses to influenza vaccination seem rarer among adults undergoing chemotherapy for cancer compared with the general population (figure).^{49,50,115-117} Vaccine responses were similar between nine patients with breast cancer and 19 healthy women, although the small size of this study means the results cannot be generalised.¹¹⁸ Serological responses among patients with solid tumours can exceed those among patients with haematological malignancies,^51,115 but not in all studies.¹¹⁹ Each of these studies had small sample sizes. Patients with lymphoma, in particular, seem to have some of the poorest humoral responses.^115,120-122 Vaccine efficacy trials with clinical infections as an outcome are lacking.

Timing of vaccination in relation to chemotherapy might be an important determinant of response. Humoral responses might be enhanced when influenza vaccination is given between chemotherapy cycles (where the blood count is lowest) rather than concurrent with chemotherapy dosing (93% vs. 50% with four-fold rise in titre to H1N1, respectively).¹¹⁹ Similarly, poor vaccine response might be related to receiving chemotherapy within 7 days before to vaccination.¹²³ No unique adverse events have been identified for influenza vaccination among patients with malignancies.

No consensus guidelines on influenza immunisation for patients with malignancies exist. Two review articles^124,125 recommended vaccination every year timed to occur either more than 2 weeks before receiving chemotherapy or between chemotherapy cycles.

Haemodialysis

By the end of 2006, 327 754 people were on chronic haemodialysis in the USA, with projected increases in prevalence.⁶ The complex and multifactorial causes of immune dysfunction in patients on haemodialysis include defects in complement activation, neutrophil function, and B-cell and T-cell function.^126-128 Infections are the second leading cause of death in patients receiving dialysis.⁶ Compared with the general population, pulmonary infection kills more patients on dialysis¹²⁹ and mortality rates from sepsis are higher.¹³⁰ However, we were unable to find any data on the specific frequencies of influenza infections and influenza-specific mortality rates in this population.

The frequency and magnitude of humoral responses to influenza vaccine among patients on dialysis can be similar to^131-136 or lower than those among people in control groups (figure).^{52-54,73,137-139} Despite potentially impaired antibody responses, a 2-year analysis of US Medicare claims data found that vaccinated patients on dialysis had a substantially lower chance of any-cause hospital admission and any-cause death than unvaccinated patients on dialysis.¹⁴⁰ This finding might indicate clinical effectiveness of vaccinating this population, but its observational design might also reflect differences in the underlying clinical status among patients vaccinated and unvaccinated. No unique adverse events related to influenza vaccine have been identified in patients on dialysis.

The CDC and authors of other reviews all recommend influenza vaccination every year for patients on haemodialysis.^18,141,142 However, only about 60% of US patients on haemodialysis were vaccinated in 2006.⁶ Many patients with end-stage renal disease also eventually undergo renal transplantation. As previously discussed, acute kidney allograft rejection can result from influenza infection.^66-68 Thus, patients on dialysis awaiting transplantation should be vaccinated to limit the risk of an influenza episode after transplantation.

Systemic corticosteroids

Many patients use systemic corticosteroids to treat a wide variety of disorders, particularly inflammatory and autoimmune disorders such as asthma, rheumatoid arthritis, and inflammatory bowel disease. Some patients receive corticosteroids for short periods, whereas others remain on chronic therapy. The prevalence of corticosteroid use is not well known. Of the total British adult population, an estimated 0·9% (or about 409 000 adult Britons) are using oral corticosteroids at any given time, with an estimated 93 000 adults using more than 7·5 mg of prednisone daily for more than 6 months.⁹ Corticosteroid use at any time was described in 2·2-9·2% of 42 500 participants in seven large cohort studies (age 55-80 years), with current use in 2-3% of participants in three of those cohorts.¹⁴³

Corticosteroids alter immune function through direct effects on gene transcription and have broad effects on non-lymphoid and lymphoid cells.¹⁴⁴ Corticosteroid use predisposes patients to infections, particularly with doses higher than 10 mg/day and cumulative doses higher than 700 mg,¹⁴⁵ but the additional risk specific for influenza posed by either chronic or short-term corticosteroid use has not been established.

Few influenza vaccination studies have been done on patients treated with corticosteroids. Four of nine studies of influenza vaccination in adult patients with rheumatoid arthritis and systemic lupus erythematosus, diseases frequently treated with systemic corticosteroids, showed impaired immune responses in patients with these rheumatological diseases compared with healthy controls,^146-149 whereas five studies found no such differences.^150-154 The power to detect small differences was limited by small sample sizes. Corticosteroids were often given in combination with other immune-modulating drugs, so the interpretation of isolated corticosteroid effects on immune response is limited. These studies also suggested that influenza vaccination does not induce a flare of the underlying rheumatological condition, though small sample sizes and limited follow-up restrict the ability to make firm conclusions on long-term safety.

Among patients with chronic lung disease, antibody responses to influenza vaccine were equivalent among 25 patients treated with corticosteroids and 14 who were not.¹⁵⁵ Similarly, humoral responses were independent of inhaled corticosteroid use (even high-dose inhaled corticosteroid use) in one study of 294 patients with asthma receiving influenza vaccine¹⁵⁶ and another of 162 patients with chronic obstructive pulmonary disease taking inhaled steroids (87 patients), oral steroids (33), and no steroids (42).¹⁵⁷

Evidence suggests that for patients taking corticosteroids—either oral or inhaled, and either chronically or for a transient period—influenza vaccination is both safe and often serologically immunogenic, but the vaccine’s clinical effectiveness in this population has not been well tested.

Limitations and challenges

Among all immunosuppressive conditions discussed, very few data exist on the prevention of clinical influenza infection by vaccination. Most studies have focused on the surrogate outcome of humoral antibody responses. However, a positive humoral response can be defined in multiple ways, most commonly a haemagglutination inhibition titre of 1:40 or greater, or a four-fold or greater rise in titres after vaccination. The limitations of these definitions become apparent, for example, in a patient who has a prevaccination titre of 1:20 that rises to 1:40. This hypothetical patient would be counted as a positive response by use of the 1:40 titre or greater criterion, but as a negative response with the four-fold rise criterion. Additionally, standard influenza vaccination includes antigens for three strains of influenza virus (H1N1, H3N2, and B). Thus, one must not only select a titre criterion, but also decide if a positive vaccine response should require titre response to one, two, or all three strains. Annual reformulation of vaccine also results in differing antigenicity of each year’s vaccine, making comparisons across studies difficult. Additionally, in some years, strains can remain unchanged from previous years. Thus, a study participant who had been vaccinated against the same strain the previous year might have a different response than a participant naive to that strain. For all these reasons, interpretation of influenza vaccine studies using humoral outcomes is challenging.

Cell-mediated immunity might also play an important part in the heightened risk of infection in the immunosuppressed populations we have reviewed. The small number of studies of cellular responses to influenza vaccination, in relatively small numbers of immunosuppressed individuals, showed impaired cellular responses among six patients¹⁵⁸ and 14 patients¹⁵⁹ after receipt of HSCTs, intact cellular responses in 65 kidney transplant recipients,¹⁶⁰ and impaired¹⁶¹ and intact¹⁶² cellular responses among lung transplant recipients. The latter study cited yearly variation in vaccine strain as one possible contributor to the discrepancy between study results in similar populations.¹⁶² Cellular immunity remains an important area for investigation, but the responses remain a surrogate outcome without clear correlations with clinical outcomes.

The best possible studies are those based on clinical outcome, but such studies with influenza vaccination in immunosuppressed individuals face substantial challenges. Unlike research on other common vaccinations (such as hepatitis B or Streptococcus pneumoniae), influenza vaccination research is faced with unpredictable yearly variation in circulating strains of virus. For example, if a study was done in a year when the circulating strain was not in that year’s vaccine, one would expect little protective effect from the vaccine, and one would likely discount any such study as flawed by factors outside of investigators’ control. Achieving adequate statistical power might also be difficult when studying the clinical effects of influenza vaccination in immunosuppressed individuals. Because most single centres lack large numbers of these patients, multicentre studies might be needed. Nevertheless, the one randomised, double-blind, controlled trial in patients with HIV showed a reduction in clinical respiratory illness, such that the result was significant with only 102 patients in the study.³³ Allocation concealment (ie, adequacy of blinding) might also be difficult when comparing placebo vaccination with influenza vaccination, which might be associated with local reactions.¹⁸ Placebo-controlled studies might also raise ethical concerns, given the large number of professional societies recommending influenza vaccination for these patients. An alternative approach would be to compare standard vaccines with newer vaccines or antiviral drugs.

We found little evidence of harm from influenza vaccination, but most studies were small and with limited follow-up. These studies had limited power to detect rare (but potentially serious) adverse events. In addition, we found almost no data on the effects of ongoing annual vaccination that most professional societies recommend (table 4).

Table 4. Recommendations for influenza vaccination of selected immunocompromised populations.

Population	Inactivated influenza vaccine recommended?	Timing of initial vaccination	Recommendations for subsequent vaccination	Sources of recommendations
HIV infection	Yes	Not specified	Annual	CDC, IDSA
Solid organ transplantation	Yes	6 months following transplantation	Annual	American Society of Transplantation
Bone marrow transplantation	Yes	Approximately 6 months following transplantation	Annual or individually assessed	CDC, IDSA, American Society of Blood and Marrow Transplantation, European Group for Blood and Marrow Transplantation
Malignancy or chemotherapy	No guidelines available	n/a	n/a	n/a
Hemodialysis	Yes	Not specified	Annual	CDC
Systemic corticosteroids	No guidelines available	n/a	n/a	n/a

CDC = US Centers for Disease Control and Prevention. IDSA = Infectious Diseases Society of America. n/a = not applicable.

Other options to control influenza infection

The priority for control of influenza is focused on generating effective antibody responses with vaccines.¹³ Progress is being made to increase the scale, duration, and breadth of vaccine responses to haemagglutinin and neuraminidase in healthy and immunocompromised populations.^163,164 Concerns that the cold-adapted live, attenuated vaccine would pose a risk to people that are immunocompromised have proven unfounded so far, but efficacy studies in adults that are immunocompromised have not been done. Driven in large part by concern about an impending influenza pandemic, areas of vaccine development include the use of standard and new adjuvants ranging from alum to toll-like receptor agonists, use of viral-like particles, other viral vectors, and DNA vaccines. Moreover, the role of more conserved antigens shared by many influenza strains, such as matrix proteins, nucleoproteins, and polymerases, might serve as targets for CD4+ and CD8+ cyototoxic T-cell responses to attenuate, if not prevent, infection and serious disease from diverse strains over time.

Chemoprophylaxis with antiviral drugs provides an immediate and immune-independent intervention to prevent seasonal influenza-related illness, as well as during family and institutional outbreaks, in both healthy and immunocompromised hosts. Indeed, chemoprophylaxis can be effective in controlling outbreaks among recipients of HSCTs.¹⁶⁵ Of equal or greater importance in this setting, are proactive and consistent infection control precautions in facilities and by employees, as proposed in ongoing plans for pandemic influenza.^166,167 Use of antiviral drugs might also help reduce mortality in patients that are immunocompromised, as suggested by an observational study of 19 recipients of HSCTs with influenza, of whom 87% received oseltamivir and none died.¹⁶⁸

Conclusions

We encourage further clinical trials with relevant clinical outcome measures, as opposed to surrogate outcomes like vaccine-induced antibody titres. We would particularly welcome randomised trials comparing standard influenza vaccine with active comparators such as modified vaccines⁵⁵ or antiviral prophylaxis with or without vaccination. Such data would greatly enhance our ability to make more informed vaccination recommendations for this population, particularly in situations of vaccine shortage or pandemic influenza.

Search strategy and selection criteria

Relevant articles were identified by searches of Medline (1966-2009) with the medical subject heading terms “influenza, human”, “influenza A virus”, “influenza B virus”, or “influenza vaccines”, combined with the following associated disease terms: “HIV”, “HIV-1”, “dialysis”, “renal dialysis”, “organ transplantation”, “transplantation”, “bone marrow transplantation”, “neoplasms”, “glucocorticoids”, “adrenal cortex hormones”, “prednisone”, and “prednisolone”. Additional relevant articles were identified by cross-referencing citations from articles identified by the Medline searches. Articles selected for inclusion reported adult influenza frequency, complications, serological or clinical responses to vaccination, or recommendations and guidelines. Because live vaccines are not recommended for immunocompromised patients, we included only articles reporting outcomes related to inactivated vaccine. Only English-language reports were included.