Seroprevalence of Pandemic Influenza Viruses, New York, New York, USA, 2004

To the Editor: Exposures to influenza viruses can lead to immune responses that substantially affect susceptibility to infection with related viruses. Characterization of preexisting immunity within a population can inform public health, as highlighted during the influenza A(H1N1)pdm09 virus pandemic, when surveillance data demonstrated that older persons (>65 years old) were less likely than younger persons to have influenza (1). Seroprevalence studies of prepandemic samples show that older persons had preexisting antibody responses to A(H1N1)pdm09 virus, presumably because of prior exposure to related strains (2). The A(H1N1)pdm09 virus possesses hemagglutinin and neuraminidase genes derived from classical swine influenza virus (3). 
 
Epidemiologic and molecular data indicate that prior exposure to early twentieth century H1N1 viruses conferred immunity to A(H1N1)pdm09 virus. Human antibodies that neutralize A(H1N1)pdm09 virus and H1N1 subtype viruses from earlier in the twentieth century have been characterized, and animal studies have demonstrated that antibodies to the earlier H1N1 subtype viruses cross-neutralize A(H1N1)pdm09 virus and protect from virus challenge (2,4–6). Prior exposure to antigenically related viruses can explain the relationship between age and susceptibility to infection. 
 
To determine the seroprevalence of preexisting hemagglutinin inhibition (HAI) antibody titers to influenza strains with pandemic potential, we tested serum samples for antibodies to A(H1N1)pdm09 virus and the 1918, 1957, and 1968 pandemic viruses. The samples had been collected in 2004 from a representative sample of adults in New York City (NYC), USA, as part of the NYC Health and Nutrition Examination Survey (Technical Appendix). For the 1918 and A(H1N1)pdm09 viruses, the highest prevalence of HAI titers >40 was among persons born before 1940 (>65 years old in 2004), although younger adults also had antibodies. Antibody prevalence to the 1957 H2N2 subtype virus was highest among persons born during 1942–1961, and >70% in persons born before 1971 had antibody to the 1968 H3N2 subtype virus (Figure). For all pandemic viruses, there was no significant difference in seroprevalence by sex or by US birth and only minor differences by race/ethnicity (Technical Appendix Table 1). 
 
 
 
Figure 
 
Seroprevalence of cross-reactive antibodies to the 1918, 1957, 1968, and 2009 pandemic influenza viruses among persons >23 years of age, New York, New York, 2004. LOESS (locally weighted scatterplot smoothing) curves represent the estimated prevalence ... 
 
 
 
We examined A(H1N1)pdm09 virus seroprevalence by the age of persons tested and by antibody titer. The mean age for persons with no serologic evidence of prior exposure (titer 40 (Technical Appendix Table 2). In a multivariate logistic regression model, presence of antibody to the 1918 H1N1 subtype virus was strongly associated with antibody to A(H1N1)pdm09 virus (Technical Appendix Table 3). No demographic factor was independently associated with positivity to A(H1N1)pdm09 virus. By using a nonlinear regression model for the probability of A(H1N1)pdm09 antibody prevalence compared with birth year, we found the model that best fit the age-stratified seroprevalence data inflected near 1927 (Technical Appendix Figure), indicating that persons born before 1927 were most reliably protected. 
 
Our findings show that the prevalence of pandemic influenza virus antibody in a representative population-based 2004 sample of NYC residents correlated with birth year and year(s) of circulating virus. These data reveal the immunologic background during the emergence of A(H1N1)pdm09 virus in NYC beginning in late April 2009 (7) and help explain why fewer cases of A(H1N1)pdm09 infection were detected among older persons than younger persons, supporting the conclusion that the difference was a result of, at least in part, antibodies elicited by prior H1N1 subtype infection in older persons. 
 
Viruses antigenically resembling the 1918 pandemic strain circulated among humans earlier in the twentieth century; cross-reactivity with antibodies to those viruses likely provided protection against the 1918 virus. Most (2,4), but not all (8), previous A(H1N1)pdm09 virus seroprevalence studies demonstrated an increase in immunity with age. In our study, more persons born before than after 1927 (i.e., persons >82 vs. those 65–82 years of age in 2009) had HAI assay results positive for A(H1N1)pdm09 virus. Protection among persons 65–82 years old during the 2009 pandemic may be explained by the presence of preexisting immunity not measured by standard HAI tests (e.g., antibodies that target the hemagglutinin stalk) or by T-cell responses (9). More positive test results were recorded with the 1918 than the A(H1N1)pdm09 virus; this finding is consistent with the model in which preexisting immunity to A(H1N1)pdm09 virus was derived from exposure to the 1918 pandemic strain or to antigenically related strains that evolved since then (10). The 1918 and 2009 strains used in testing may have exhibited different sensitivities in HAI assays. Immunity in older populations is not surprising and was seen in the 1918, 1957, and 1968 pandemics, during which newly introduced pandemic viruses were more likely to cause clinical illness in younger persons, presumably because prior exposure to similar viruses resulted in cross-reactive antibodies (11). 
 
Study limitations include a relatively small sample size and a lack of history regarding influenza virus infection or vaccination. Nevertheless, the ability to evaluate seroreactivity in a representative sample of adults helps validate and reinforce previously published findings on H1N1 subtype viruses and clarifies levels of immunity to H2N2 and H3N2 subtype viruses. 
 
Technical Appendix: 
Information on datasets and methods, the characteristics of persons with cross-reactive antibodies to pandemic influenza viruses and different antibody titers to influenza A(H1N1)pdm09 virus, estimates of logistic regression parameters, and the predicted probability of cross-reactivity to influenza A(H1N1)pdm09 virus. 
 
Click here to view.(315K, pdf)


Seroprevalence of Pandemic Infl uenza
Viruses, New York, New York, USA, 2004 To the Editor: Exposures to infl uenza viruses can lead to immune responses that substantially affect susceptibility to infection with related viruses. Characterization of preexisting immunity within a population can inform public health, as highlighted during the infl uenza A(H1N1)pdm09 virus pandemic, when surveillance data demonstrated that older persons (>65 years old) were less likely than younger persons to have infl uenza (1). Seroprevalence studies of prepandemic samples show that older persons had preexisting antibody responses to A(H1N1) pdm09 virus, presumably because of prior exposure to related strains (2). The A(H1N1)pdm09 virus possesses hemagglutinin and neuraminidase genes derived from classical swine infl uenza virus (3).
Epidemiologic and molecular data indicate that prior exposure to early twentieth century H1N1 viruses conferred immunity to A(H1N1) pdm09 virus. Human antibodies that neutralize A(H1N1)pdm09 virus and H1N1 subtype viruses from earlier in the twentieth century have been characterized, and animal studies have demonstrated that antibodies to the earlier H1N1 subtype viruses crossneutralize A(H1N1)pdm09 virus and protect from virus challenge (2,(4)(5)(6). Prior exposure to antigenically related viruses can explain the relationship between age and susceptibility to infection.
To determine the seroprevalence of preexisting hemagglutinin inhibition (HAI) antibody titers to infl uenza strains with pandemic potential, we tested serum samples for antibodies to A(H1N1)pdm09 virus and the 1918, 1957, and 1968 pandemic viruses. LETTERS The samples had been collected in 2004 from a representative sample of adults in New York City (NYC), USA, as part of the NYC Health and Nutrition Examination Survey (online Technical Appendix, wwwnc.cdc. gov/EID/pdfs/12-0156-Techapp.pdf). For the 1918 and A(H1N1)pdm09 viruses, the highest prevalence of HAI titers >40 was among persons born before 1940 (>65 years old in 2004), although younger adults also had antibodies. Antibody prevalence to the 1957 H2N2 subtype virus was highest among persons born during 1942-1961, and >70% in persons born before 1971 had antibody to the 1968 H3N2 subtype virus ( Figure). For all pandemic viruses, there was no signifi cant difference in seroprevalence by sex or by US birth and only minor differences by race/ ethnicity (online Technical Appendix Table 1).
We examined A(H1N1)pdm09 virus seroprevalence by the age of persons tested and by antibody titer. The mean age for persons with no serologic evidence of prior exposure (titer <20) was 50 years, compared with 72 years for those with titers of 20-40 and 80 years for those with titers >40 (online Technical Appendix Table 2). In a multivariate logistic regression model, presence of antibody to the 1918 H1N1 subtype virus was strongly associated with antibody to A(H1N1)pdm09 virus (online Technical Appendix Table 3). No demographic factor was independently associated with positivity to A(H1N1) pdm09 virus. By using a nonlinear regression model for the probability of A(H1N1)pdm09 antibody prevalence compared with birth year, we found the model that best fi t the age-stratifi ed seroprevalence data infl ected near 1927 (online Technical Appendix Figure), indicating that persons born before 1927 were most reliably protected.
Our fi ndings show that the prevalence of pandemic infl uenza virus antibody in a representative population-based 2004 sample of NYC residents correlated with birth year and year(s) of circulating virus. These data reveal the immunologic background during the emergence of A(H1N1) pdm09 virus in NYC beginning in late April 2009 (7) and help explain why fewer cases of A(H1N1)pdm09 infection were detected among older persons than younger persons, supporting the conclusion that the difference was a result of, at least in part, antibodies elicited by prior H1N1 subtype infection in older persons.
Viruses antigenically resembling the 1918 pandemic strain circulated among humans earlier in the twentieth century; cross-reactivity with antibodies to those viruses likely provided protection against the 1918 virus. Most (2,4), but not all (8), previous A(H1N1)pdm09 virus seroprevalence studies demonstrated an increase in immunity with age. In our study, more persons born before than after 1927 (i.e., persons >82 vs. those 65-82 years of age in 2009) had HAI assay results positive for A(H1N1)pdm09 virus. Protection among persons 65-82 years old during the 2009 pandemic may be explained by the presence of preexisting immunity not measured by standard HAI tests (e.g., antibodies that target the hemagglutinin stalk) or by T-cell responses (9). More positive test results were recorded with the 1918 than the A(H1N1)pdm09 virus; this fi nding is consistent with the model in which preexisting immunity to A(H1N1)pdm09 virus was derived from exposure to the 1918 pandemic strain or to antigenically related strains that evolved since then (10). The 1918 and 2009 strains used in testing may have exhibited different sensitivities in HAI assays. Immunity in older populations is not surprising and was seen in the 1918, 1957, and 1968 pandemics, during which newly introduced pandemic viruses were more likely to cause clinical illness in younger persons, presumably because prior exposure to similar viruses resulted in cross-reactive antibodies (11).
Study limitations include a relatively small sample size and a lack of history regarding infl uenza virus infection or vaccination. Nevertheless, the ability to evaluate seroreactivity in a representative sample of adults helps validate and reinforce previously published fi ndings on H1N1 subtype viruses and clarifi es levels of immunity to H2N2 and H3N2 subtype viruses.
This study was supported, in part, by National Institutes of Health grants AI072258 and U54 AI057158 (the Northeast Biodefense Center-Lipkin grant) to C.F.B.  To the Editor: Streptomyces spp. are aerobic, gram-positive bacteria of the order Actinomycetales, known for their ability to produce antimicrobial molecules such as streptomycin. Streptomyces spp., usually saprophytic to humans, can cause local cutaneous fi stulized nodules known as actinomycetoma or mycetoma. Severe invasive infections have seldom been reported, but most cases reported have occurred in immunocompromised patients (1)(2)(3)(4)(5). We report a case of invasive pulmonary infection caused by a Streptomyces sp. in a splenectomized patient with sarcoidosis.
In 2003, multiorgan sarcoidosis was diagnosed in a man, 57 years of age; the disease involved lungs, skin, joints, and lymph nodes. Corticosteroids were initially given but quickly discontinued because of a severe psychiatric reaction. In 2007, a splenectomy was performed on this patient to remove an intestinal obstruction caused by a severely enlarged spleen, identifi ed as a specifi c localization of sarcoidosis.
In April 2008, the patient was admitted to the internal medicine unit of Saint-André Hospital in Bordeaux, France with fever (38.9°C/102°F), progressive asthenia, anorexia, weight loss, productive cough, and New York Heart Association grade III dyspnea. Bilateral basal crackles could be heard in the lungs; physical examination fi ndings were otherwise within normal limits. Biological tests showed infl ammatory syndrome with elevated C-reactive protein (74 mg/L, reference value <5 mg/L) without any other consequential abnormality. Gamma globulin levels were normal. A chest radiograph showed bilateral interstitial infi ltrate. A computed tomogra-

NYC Health and Nutrition Examination Survey
In 2004, the New York City Department of Health and Mental Hygiene (NYCDOHMH) conducted a population-based, cross sectional survey of NYC adults modeled after the National Health and Nutrition Examination Survey (HANES). The methodology for NYC HANES has been described in detail elsewhere (1). In brief, the survey population included a representative sample of non-institutionalized adults, ages 20 and older, recruited through letters and field visits, followed up by interviews, physical examination and biologic specimen collection for consenting participants. A 3-stage cluster sampling plan was used to recruit participants between June and December 2004. The stages of sample selection were as follows: 1) selection of census blocks or groups of blocks; 2) random selection of households within selected segments; and 3) random selection of study participants within households. The survey included a face-to-face computer-assisted personal interview, private audio computer-assisted self-interview, physical examination, and laboratory testing (sera and urine). The overall response rate was 55%.
Participants were asked to consent to allow for collection and storage of additional biologic samples for future research purposes, without notification of the results. Demographic data collected at the time of interview included age, race/ethnicity, country of birth, and gender.
Occupational and influenza histories were not collected.

Specimen Sampling
From the specimen repository we randomly selected serum samples from those persons born in 1981 or earlier (persons age 23 or older at the time of the NYC HANES collection) to include in this study. This would allow us to better understand prior exposure to H1N1 (presumably conferring immunity), which might explain the age-related pattern of illness seen The subsample was drawn in such a way as to address two distinct purposes of the study.
The first was to establish a cutoff birth date, before which individuals may show an increased likelihood of cross-reactivity to the 2009 H1N1 virus . The second was to estimate the level of cross-reactive antibodies for those born before the cutoff. The first purpose favors selecting more samples around the potential cutoff while the second favors selecting sample from respondents born before the cutoff. After establishing a potential range for the cutoff value, the sample was drawn with increasing probability in the range, very low probability below the range, and moderate probability above the range. Component weights were adjusted by their inverse probability of selection into the subsample and then scaled to the population total.

Laboratory Testing
Human Sera Treatment: Human sera was treated with trypsin periodate, according to established procedures, to remove nonspecific inhibitors of hemagglutination (2).

Virus and Virus like particle (VLP) production:
The 1918 pandemic H1N1 and 1957 pandemic H2N2 viruses were assayed using noninfectious virus-like particles (VLPs) rather than live virus to allow for use at biosafety level 1 (BSL1). In brief, VLPs were produced by co-transfecting HEK293T cells with expression

Statistical Analysis
For all analyses a hemagglutination titer of 40 or greater was used to indicate prior exposure and/or immunity to the virus in question. Laboratory data were merged with demographic information including date of birth, age (as of 2004), gender, country of birth (USA vs. foreign born), and race/ethnicity to determine the prevalence of prior exposure by these factors.
Locally weighting smoothing scatterplots (LOESS curves) (5) were used to estimate the prevalence of antibodies by year of birth. To appropriately account for the complex survey design of NYC HANES, logistic regression models were fit using Sudaan version 10.0 (6).
Nonlinear models were used to explicitly estimate the cutoff year for increased probability of immunity, using a segmented regression model that includes cutoff year as a parameter (7).   compared with year of birth, we found that the model that best fits the age-stratified seroprevalence data inflects near birth year of 1927, suggesting a cut-off point for immunity as measured by hemagglutination inhibition assay estimated as July 3, 1927, (95% confidence limits: May 2, 1925, September 3, 1929, and for those after, there is little detectable immunity.