Vaccine 28 (2010) 2860–2864 Contents lists available at ScienceDirect Vaccine journal homepage: www.elsevier.com/locate/vaccine Improved anamnestic response among adolescents boosted with a higher dose of the hepatitis B vaccine夽 Sandra S. Chaves a,∗ , Justina Groeger a , Louisa Helgenberger b , Steven B. Auerbach c , Stephanie R. Bialek a , Dale J. Hu a , Jan Drobeniuc a a b c Division of Viral Hepatitis, Centers for Disease Control & Prevention, United States Department of Health & Social Affairs, Federated States of Micronesia National Government, Palikir, Pohnpei, Federated States of Micronesia New York Regional Division, Office of the Performance Review, Health Resources and Services Administration, United States a r t i c l e i n f o Article history: Received 20 November 2009 Received in revised form 22 January 2010 Accepted 28 January 2010 Available online 12 February 2010 Keywords: Anamnestic response Hepatitis B vaccination Hepatitis B virus Long term immunity a b s t r a c t Some hepatitis B vaccine booster studies have suggested waning of vaccine-induced immunity in adolescents vaccinated starting at birth. Those studies, however, used a pediatric formulation of the hepatitis B vaccine as a booster to detect anamnestic response. We compared adolescents boosted with an adult dose of hepatitis B vaccine with those boosted with a pediatric dose. Among adolescents who had lost protective antibody levels against hepatitis B, a higher proportion had an anamnestic response when boosted with the adult dose (60.0% vs. 43.8%). Thus, higher antigen concentrations may be required to elicit an adequate immune memory response. Despite improved anamnestic response, our study still raises concerns about whether children immunized in early infancy will remain protected from hepatitis B as they age into adulthood. Published by Elsevier Ltd. 1. Introduction Three or more doses of hepatitis B vaccine, with first dose given at birth, have been shown to induce protective antibody responses in ≥96% of vaccinated infants [1]. However, the relative immaturity of the infant immune system during primary vaccination may preclude an optimal immune memory response, and, subsequently, the likelihood of achieving long term protection against hepatitis B virus (HBV) infection [2–4]. Although levels of antibody to hepatitis B surface antigen (anti-HBs) decline rapidly within the first year after vaccination, and progressively in subsequent years, diminished antibody titers do not necessarily correlate with lack of protection [5]. To demonstrate lasting vaccine-induced immunity long after anti-HBs becomes undetectable, a booster dose of the vaccine can be given to previously vaccinated individuals. A vigorous increase in anti-HBs titers detected shortly after giving an extra dose of the vaccine can be interpreted as an anamnestic response due to preserved immune memory [2,3,6–9]. In two recent studies, one among Alaska Natives [7] and one conducted in Micronesia [10], a pediatric dose of recombinant hepatitis B vaccine was administered to adolescents ∼15 years after completion of the primary vaccine series. In both studies, half of the participants failed to generate an anamnestic response, raising concerns about whether the current hepatitis B vaccination schedule is adequate to ensure protection against HBV infection for children immunized in early infancy as they age into adulthood. Because adolescents are constitutionally more similar to adults than young children, we hypothesized that some adolescents may require a higher amount of antigen to mount an adequate anamnestic response. To validate our hypothesis, we boosted a group of adolescents with an adult dose of hepatitis B vaccine and compared their anamnestic response to those of adolescents boosted with a pediatric dose of the same vaccine. 2. Materials and methods 2.1. Study setting and population 夽 The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention, US Department of Health and Human Services. ∗ Corresponding author at: Centers for Disease Control & Prevention, 1600 Clifton Road, NE; MS G-37, Atlanta, GA 30333, United States. Tel.: +1 404 718 8551; fax: +1 404 718 8585. E-mail address: schaves@cdc.gov (S.S. Chaves). 0264-410X/$ – see front matter. Published by Elsevier Ltd. doi:10.1016/j.vaccine.2010.01.059 This study was conducted in Pohnpei, Federated States of Micronesia (FSM), where hepatitis B is of intermediate endemicity. Recombinant hepatic B vaccine (Recombivax, Merck and Co., Whitehouse Station, NJ) was first used in FSM in 1989 and incorporated into the routine infant immunization program in 1990. The licensed dosing schedule during the time that study participants S.S. Chaves et al. / Vaccine 28 (2010) 2860–2864 were vaccinated was 5 ␮g at birth and 2.5 ␮g at 2 and 6 months of age. We drew our study sample from a cohort of adolescents who were vaccinated during 1989–1991 and were part of a serosurvey conducted by the FSM Department of Health Services in 1993 [11]. As a part of that survey, health department employees visited a stratified random sample of households on the main and neighboring islands to collect detailed information on vaccination and serum samples to be tested for markers of HBV infection. In 2006, a subset of that original cohort of subjects was enrolled in a 15-year follow-up study on the persistence of hepatitis B vaccine-induced immunity—they were given a pediatric-booster dose of hepatitis B vaccine; half had anamnestic response [10]. In 2008, we were able to recruit a comparison group of adolescents for our study among that same original cohort. For our study, subjects were eligible if had tested negative to hepatitis B core antigen (anti-HBc), received three doses of hepatitis B vaccine beginning at birth, and were not boosted as part of the 2006 pediatric-booster dose study [10]. 2.2. Study design This study was approved by the CDC Institutional Review Board and the FSM Department of Health and Social Services. After written consent, a serum sample was collected from each participant and an adult dose of hepatitis B vaccine (Recombivax® , 10 mcg) was given to each participant. Fourteen days later, another serum sample was collected to assess the presence of anamnestic response, and therefore, estimate the duration of vaccine-induced protection. Protection against HBV infection was defined as titer of anti-HBs ≥10 mIU/mL. An anamnestic response was defined as a 4-fold increase in anti-HB concentration after receipt of a booster dose, with a peak titer of anti-HBs ≥10 mIU/mL. Chronic HBV infection was defined as a positive test for HBsAg. 2.3. Laboratory Serum samples were stored frozen, shipped to the CDC Hepatitis Reference Laboratory in Atlanta, and tested for anti-HBc (VITROS aHBc), HBsAg (VITROS HBsAg), and anti-HBs (VITROS aHBs, OrthoClinical Diagnostics, Inc., Rochester, NY). Anti-HB concentrations were determined in mIU/mL. 2.4. Analysis Geometric mean concentrations (GMCs) were calculated for anti-HBs titers. Participants with no detectable anti-HBs were assigned a value of 0.05 mIU/mL for the calculation of GMC. Data were analyzed using SAS 9.1 (Cary, NC). The Chi-square test was used to compare categorical variables and the Wilcoxon rank sum 2861 test was used to compare continuous variables. The Kruskal–Wallis test was used to compare GMCs between groups. 3. Results From a total of 158 eligible participants, 89 adolescents were enrolled in the adult-booster dose comparison study group. There were no differences between eligible participants and non-participants in the 2008 adult-booster dose study (Table 1). The median age (range) of the pediatric-booster dose study group was 15.7 years (14.9, 16.8) compared to 17.7 years (16.7, 18.7) from our adult-booster study group (p = <0.001) (Table 2). The median age of participants in our study at the time of completion of primary vaccination series was 10.5 months whereas it was 6.8 months for those participants in the 2006 pediatricbooster dose study. A higher percentage of the latter group received a birth dose (86.7% compared to 74.2% in our study; p = 0.03). Among the 89 participants enrolled in the adult-booster study, 14 (16%) had become anti-HBc positive since the 1993 baseline testing. None of those 14 anti-HBc positive cases were chronically infected, and 6 had anti-HBs titer <10 mIU/mL. For the remaining 75, we did not have pre-booster anti-HBs titer for one of the boosted participants. Among anti-HBc negative participants, 9% (7/74) retained protective antibody levels (anti-HBs titer ≥10 mIU/mL) pre-booster compared to 7% [10] observed in the pediatric-booster dose study (p = 0.60). We further excluded 7 participants from the post-booster dose analysis as they did not get the 14-day post-booster dose blood draw, resulting in a total of 67 participants. Of the remaining 67 participants, 60 had anti-HBs titers below protective levels and of those, 60.0% (36/60) had an anamnestic response when challenged with the adult-booster dose. In comparison, 43.8% (39/89) of those adolescents challenged with a pediatric-booster dose of the hepatitis B vaccine had an anamnestic response (p = 0.05). When we account for all 68 participants with blood drawn postbooster dose (which includes the one participant who did not have pre-booster anti-HBs titer but post-booster titer was 102 mIU/mL), 65% (95% CI 52–76%) could be considered protected based on levels of anti-HBs titers ≥10 mIU/mL. Overall, after boosting, there were 35% of adolescents with anti-HBs titer below protection levels among the adult-booster study compared to 52% among those in the pediatric-booster study (p = 0.03). Comparison of anti-HBs preand post-boosting by study groups can be seen in Fig. 1. GMC post-booster was higher among those who had protective anti-HBs levels before being boosted (2076.5 compared to 81.1 among those with <10 mIU/mL before booster). The overall postbooster GMC among anamnestic responders in our study was 128.9 which was not statistically significantly different from those in Table 1 Characteristics of eligible adolescents participants and non-participants in the adult-booster dose study, Pohnpei, Federate States of Micronesia—2008. Characteristics 2008 eligible N = 158 Male Age at time of primary hepatitis B vaccine series Dose 1, median age in days (range) Dose 2, median age in months (range) Dose 3, median age in months (range) First dose given within 2 days of birth Median age in months at 1993 baseline study survey (range) Median time in years since vaccination at 2008 follow-up study (range) Anti-HBs ≥10 mIU/mL at 1993 baseline studya a b 82 0 1.6 8.8 117 35.0 16.7 72b 2008 participants 2008 non-participants N = 89 N = 69 (51.9%) 46 (51.7%) (0–1296) (1.1–45.3) (3.4–52.4) 0 1.6 10.5 (0–528) (1.2–28.9) (3.9–52.4) (74.1%) (24.6–47.8) (14.3–18.3) (55.8%) 66 36.2 16.6 40b (74.2%) (24.6–47.8) (14.3–18.3) (53.3%) Median time since completion of vaccination series ∼24 months. Percentage calculated based on available data—29, 14 and 15 missing information respectively. 36 0 1.6 7.6 51 33.1 16.7 32b p-Value (52.2%) 0.95 (0–1296) (1.1–45.3) (3.4–49.7) 0.88 0.57 0.16 (73.9%) (24.6–46.7) (14.3–18.2) (59.3%) 0.97 0.23 0.81 0.50 2862 S.S. Chaves et al. / Vaccine 28 (2010) 2860–2864 Table 2 Characteristics of adolescents participants in the pediatric and adult-booster dose studies, Pohnpei, Federate States of Micronesia—2006 and 2008. Characteristics 2008 participants 2006 participants Adult-booster study Pediatric-booster study N = 89 N = 105 Male 46 (51.7%) Age at time of primary hepatitis B vaccine series Dose 1, median age in days (range) Dose 2, median age in months (range) Dose 3, median age in months (range) 0 1.6 10.5 (0–528) (1.2–28.9) (3.9–52.4) Median age in years at time of follow-up study (range) First dose given within 2 days of birth Median age in months at 1993 baseline study (range) Anti-HBs ≥10 mIU/mL at 1993 baseline studya 17.7 66 36.2 40b (16.7, 18.7) (74.2%) (24.6–47.8) (53.3%) a b p-Value 48 0 1.51 6.8 15.7 91 34.8 42 (45.7%) 0.41 (0–423) (1.2–29.0) (3.4–46.6) 0.03 0.05 0.002 (14.9–16.8) (86.7%) (24.9–47.8) (40.0%) <0.001 0.03 0.50 0.08 Median time since completion of vaccination series for both study groups was 25.14 months. Percentage calculated based on available data—14 missing information. Table 3 Level of antibody to hepatitis B surface antigen (anti-HBs) and the geometric mean concentration (GMC) before and after receipt of a booster dose of hepatitis B vaccine among anamnestic responders in the pediatric and adult-booster dose studies—Pohnpei, Federate States of Micronesia—2006 and 2008. Study group Pre-booster anti-HBs (mIU/mL) Pre-booster GMC (range) Post-booster GMC (range)a Pediatric-booster (2006) <10 (n = 39) ≥10 (n = 6) Total (n = 45) 0.2 (0, 9.4) 89.0 (28.3, 946.4) 0.4 (0, 946.4) 152.6 (10.6, 2944.3) 2689.8 (549.3, 17134) 223.7 (10.6, 17134) Adult-booster (2008) <10 (n = 36) ≥10 (n = 6) Total (n = 42) 0.8 (0, 5.6) 40.3 (20.6, 89.8) 1.4 (0, 89.8) 81.1 (10.1, 3720) 2076.5 (109, 94300) 128.9 (10.1, 94300) a Kruskal–Wallis test comparing total post-booster GMC for adolescents in the adult dose study (GMC = 128.9) with that of those in the pediatric dose study (GMC = 223.7); p = 0.10. Fig. 1. Comparison of anti-HB concentrations pre- and post-booster dose among all adolescents boosted with a pediatric dose (n = 96) of the hepatitis B vaccine and those boosted with an adult dose (n = 68), Pohnpei, 2006 and 2008. the pediatric-booster dose study, where GMC was 223.7 (p = 0.10) (Table 3). 4. Discussion In our study, by using an adult dose of the hepatitis B vaccine, we demonstrated that adolescents who had previously lost protective levels of anti-HBs were more likely to mount an immune response than adolescents who were boosted with a pediatric dose of the vaccine (anamnestic response 60.0% vs. 43.8%, respectively). Thus, a higher hepatitis B antigen concentration might be needed to trigger an optimal immune memory response long after anti-HBs titers become undetectable. This is the first study to investigate whether the capacity to generate an adequate anamnestic response among those vaccinated starting at birth is associated with the antigen concentration used for the booster dose given. Dentico et al. [12] have described a dosedependent immune response to different antigen concentrations given to healthy vaccine-responders. The participants in that study, however, were vaccinated as adults and were challenged using nonadsorbed HBsAg to mimic natural exposure to the wild-type HBV. The amplitude of the response to the challenging antigen concentration received was also dose-dependent; moreover, it translated into higher GMCs as antigen-dose increased. Contrary to what we would expect, in our study the GMCs among those receiving a higher booster dose were comparable to that of those boosted with a pediatric dose of the vaccine—perhaps because pre-boosting GMCs were very low (GMC 1.4 mIU/mL to start with). A 22-year follow-up study among Alaska Natives vaccinated after 6 months of age described increased GMCs after boosting [13], similar to that described by Dentico et al. The participants in the Alaska study, however, had a higher pre-booster anti-HBs GMC (21.5 mIU/mL) than that seen in our study. There is a clear association between the anti-HBs titers observed pre-boosting and the magnitude of anamnestic response observed after boosting [13,14]; and strong immunological memory has also been shown to persist among adolescents immunized at an older age during infancy, i.e., vaccination starting at 3 months or later [13–15]. The relative immaturity of the infant immune system could explain why the immune memory might be suboptimal among persons vaccinated beginning at birth compared with cohorts vaccinated at an older age and could explain the low GMCs observed ∼15–17 years after primary vaccination. Following the primary immune response to hepatitis B vaccination, both memory T cells and memory B cells are generated and contribute to subsequent antibody responses trigged by reexposure to the virus antigen. Among Alaska Natives vaccinated with recombinant hepatitis B vaccine with first dose given at birth, available data show a gradual decline in anamnestic response over time; the rates of anamnestic response 5, 10 and 15 years after completion of primary vaccination were 97% [7], 79% [7], and 51% [8] respectively. Similar decline in percentage of partic- S.S. Chaves et al. / Vaccine 28 (2010) 2860–2864 ipants with anamnestic response are seen in studies conducted in the US-affiliated Pacific Islands; at 5-year follow-up [16], 100% of vaccinated children responded to a booster dose, with 81% [17] and 47% [10] of anamnestic response rate seen at 10- and 15-year follow-up studies respectively. Even with improved response rate as described in our study, trends suggest waning of immunity. This could pose a problem for countries relying on early infancy hepatitis B vaccination, like the United States, where the risk of HBV infection increases again during adolescence and early adulthood because of sexual or behavioral exposures. There were 16% anti-HBc positive adolescents in our study which was comparable to that reported in the pediatric dose study [10], where 7.6% of fully vaccinated adolescents were anti-HBc positive (p = 0.08). In a 10-year follow-up study done in Palau [17] and Taiwan [18], highly hepatitis B endemic countries, the rates of antiHBc positivity among fully vaccinated children 10 years later were 7% and 15% respectively. The rate of breakthrough infection in our study was also similar to that of a follow-up study done in The Gambia, where 17.7% of adolescents fully vaccinated against hepatitis B in infancy were anti-HBc positive 15 years later [19]. Breakthrough hepatitis B infection shown in most of those studies was not accompanied by detection of chronic infection but could corroborate waning of vaccine-induced immunity. In our study, most anti-HBc positive participants likely represent resolved breakthrough infections, although 6 had levels of anti-HBs titers <10 mIU/mL which could indicate an ‘occult’ chronic infection (but specimens were not tested for viral load). A recent study has documented perinatal transmission of HBV in infants born to anti-HBc positive but HBsAg negative mothers [20]. Although none of the infants in that study developed chronic HBV infection, long term clinical implications are unknown. Limitations in our study would be related to potential differences among the two comparison groups. If our study group were exposed to HBV more often than those in the comparison group, for instance, a differential effect of natural boosting could occur. However, as the percentage of those with anti-HBs levels ≥10 mIU/mL was comparable in both groups pre-boosting, this possibility seems unlikely. The adult-booster dose study group was slightly older at completion of primary vaccination than those receiving pediatric dose [10]. To address this issue, we estimated anamnestic response for a sub-group of adolescents who had completed primary vaccination before 7 months of age. Anamnestic response for this sub-group of adolescents was 57% (12/21), still higher than that seen in the pediatric-boosted group (47%), though due to the small number of observations, the difference was no longer statistically significant. Another difference between the two groups is that another 2 more years since primary vaccination had elapsed between the two studies, which could have biased our results to the null. Despite more time having elapsed since primary vaccination, a greater proportion of our study participants were able to mount an anamnestic response after being challenged with an adult dose of the vaccine. Conceivably, if the two booster studies had been done simultaneously, the group receiving the adult-booster dose of the hepatitis B vaccine could have had an even higher percentage with anamnestic response. Moreover, the role of anamnestic response as proxy measure for duration of vaccine-induced immunity is not fully understood. Prompt antibody response is fundamental to neutralize HBV and stop the progression of infection. Therefore, if immune memory fails over time to generate anti-HBs after exposure to the virus antigen (i.e., as seen by the lack of anamnestic response), individuals vaccinated starting at birth may be susceptible to HBV infection if exposed to the virus during adolescence and adulthood. Despite improved anamnestic responses with a higher dose of the hepatitis B vaccine, the overall 65% protection rate among our study participants is still concerning. Whether exposure to wild-type HBV 2863 could have elicited a more powerful immune response to halt infection is yet not known. Further assessment of dose response to HBV antigen is warranted to better understand the duration of protection among those vaccinated starting at birth as this has important policy implications in determining whether or not a booster vaccination is needed. Acknowledgments We are extremely grateful to all participants and those who contributed to the implementation of this project—the field study team and nurses, in particular Mr Spencer Donre, Mr Hinden Allexander and Ms Kiomy Noket, and CDC Hepatitis Reference Laboratory personnel—Ms Tracy Green-Monthfort and Mrs Ngok-Thao Le. This study could not be done without the valuable support of the Department of Health & Social Affairs, Federated States of Micronesia National Government, Palikir, Pohnpei, FSM. This research was supported in part by an appointment to the Research Participation Program at the Centers for Disease Control and Prevention administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the U.S. Department of Energy and CDC. References [1] Centers for Disease Control and Prevention. A comprehensive immunization strategy to eliminate transmission of hepatitis B virus infection in the United States: recommendations of the Advisory Committee on Immunization Practices (ACIP). Part 1. Immunization of infants, children, and adolescents. MMWR Recomm Rep 2005;54(RR-16):1–31. [2] Banatvala J, Van Damme P, Oehen S. Lifelong protection against hepatitis B: the role of vaccine immunogenicity in immune memory. Vaccine 2000;19:877–85. [3] Petersen KM, Bulkow LR, McMahon BJ, Zanis C, Getty M, Peters H, et al. Duration of hepatitis B immunity in low risk children receiving hepatitis B vaccinations from birth. Pediatr Infect Dis J 2004;23:650–5. [4] Lee SS, Young BW, Wong KH, Lim WL. The implication of a reduced-dose hepatitis B vaccination schedule in low risk newborns. Vaccine 2002;20: 3752–4. [5] Dentinger CM, McMahon BJ, Butler JC, Dunaway CE, Zanis CL, Bulkow LR, et al. Persistence of antibody to hepatitis B and protection from disease among Alaska natives immunized at birth. Pediatr Infect Dis J 2005;24:786–92. [6] Zanetti AR, Mariano A, Romanò L, D’Amelio R, Chironna M, Coppola RC, et al. Long-term immunogenicity of hepatitis B vaccination and policy for booster: an Italian multicentre study. Lancet 2005;366:1379–84. [7] Samandari T, Fiore AE, Negus S, Williams JL, Kuhnert W, McMahon BJ, et al. Differences in response to a hepatitis B vaccine booster dose among Alaskan children and adolescents vaccinated during infancy. Pediatrics 2007;120:e373–81. [8] Hammitt LL, Hennessy TW, Fiore AE, Zanis C, Hummel KB, Dunaway E, et al. Hepatitis B immunity in children vaccinated with recombinant hepatitis B vaccine beginning at birth: a follow-up study at 15 years. Vaccine 2007;25:6958–64. [9] Taiwan study 15 years f/up and CMI. [10] Bialek SR, Bower WA, Novak R, Helgenberger L, Auerbach SB, Williams IT, et al. Persistence of protection against hepatitis B virus infection among adolescents vaccinated with recombinant hepatitis B vaccine beginning at birth: a 15-year follow-up study. Pediatr Infect Dis J 2008;27(October (10)):881–5. [11] Mahoney FJ, Woodruff BA, Erben JJ, Coleman PJ, Reid EC, Schatz GC, et al. Progress on the elimination of hepatitis B virus transmission in Micronesia and American Samoa. Pac Health Dialog 1996;3:140–6. [12] Dentico P, Crovari P, Lai PL, Ponzio F, Safary A, Pellegrino A, et al. Anamnestic response to administration of purified non-adsorbed hepatitis B surface antigen in healthy responders to hepatitis B vaccine with long-term non-protective antibody titres. Vaccine 2002;20(November (31–32)):3725–30. [13] McMahon BJ, Dentinger CM, Bruden D, Zanis C, Peters H, Hurlburt D, et al. Antibody levels and protection after hepatitis B vaccine: results of a 22-year follow-up study and response to a booster dose. J Infect Dis 2009;200(November (9)):1390–6. [14] Zanetti AR, Mariano A, Romanò L, D’Amelio R, Chironna M, Coppola RC, et al. Long-term immunogenicity of hepatitis B vaccination and policy for booster: an Italian multicentre study. Lancet 2005;366(October (9494)):1379–84. [15] Gabbuti A, Romanò L, Blanc P, Meacci F, Amendola A, Mele A, et al. Longterm immunogenicity of hepatitis B vaccination in a cohort of Italian healthy adolescents. Vaccine 2007;25(April (16)):3129–32. Epub 2007 Jan 19. [16] Williams IT, Goldstein ST, Tufa J, Tauillii S, Margolis HS, Mahoney FJ. Long term antibody response to hepatitis B vaccination beginning at birth and to subsequent booster vaccination. Pediatr Infect Dis J 2003;22:157–63. [17] Patel P, Armstrong G, Samandari T, Talekar S, Yamanguchi K, Kuhnert W, et al. Long term effectiveness of hepatitis B vaccine in Palau. In: Presented at: 2864 S.S. Chaves et al. / Vaccine 28 (2010) 2860–2864 abstracts of the infectious diseases society of American 43rd annual meeting. 2004. Abstract. [18] Wu JS, Hwang LY, Goodman KJ, Beasley RP, Hepatitis. B vaccination in high-risk infants: 10-year follow-up. J Infect Dis 1999;179(June (6)):1319–25. [19] van der Sande MA, Waight PA, Mendy M, Zaman S, Kaye S, Sam O, et al. Long-term protection against HBV chronic carriage of Gambian adolescents vaccinated in infancy and immune response in HBV booster trial in adolescence. PLoS One 2007;2(August (1)):e753. [20] Walz A, Wirth S, Hucke J, Gerner P. Vertical transmission of hepatitis B virus (HBV) from mothers negative for HBV surface antigen and positive for antibody to HBV core antigen. J Infect Dis 2009;200(October (8)):1227–31.