Long-term follow-up of peginterferon and lamivudine combination treatment in HBeAg-positive chronic hepatitis B

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

We have previously demonstrated that combination peginterferon and lamivudine treatment has superior antiviral efficacy to lamivudine monotherapy in chronic hepatitis B. In this study, we investigated the long-term posttreatment virological response to this combination treatment. Sustained virological response of patients who completed 32-week peginterferon and 52-week lamivudine combination treatment was compared to patients who completed 52-week lamivudine monotherapy. Sustained response was defined as sustained hepatitis B e antigen (HBeAg) loss and HBV DNA < 100,000 copies/mL from treatment cessation until the end of follow-up. Forty-eight patients receiving combination treatment and 47 patients receiving lamivudine monotherapy were studied. The posttreatment follow-up of patients who received combination treatment was 117 ± 34 weeks and that of patients receiving lamivudine monotherapy was 124 ± 29 weeks. At the end of treatment, HBeAg loss occurred in 63% of patients in the combination group and 28% of patients in the lamivudine group (P = .001). The probabilities of sustained response for combination treatment and lamivudine monotherapy were 33% and 13% at week 24, 31% and 11% at week 52, and 29% and 9% at week 76, respectively (log-rank test, P = .0015). No patients developed virological relapse after week 76 until the last visit in either treatment group. All sustained responders had no biochemical relapse (alanine aminotransferase [ALT] > 2 times upper limit of normal) during follow-up. Among the non-sustained responders, biochemical relapse occurred in 32 patients (94%) in the combination group and 38 patients (88%) in the lamivudine group, respectively. In conclusion, combination treatment of peginterferon and lamivudine has a higher sustained virological response than lamivudine monotherapy up to 3 years after treatment. (HEPATOLOGY 2005.)

Chronic hepatitis B is the major cause of liver cirrhosis and hepatocellular carcinoma in Asia. Patients who have positive hepatitis B e antigen (HBeAg) and active liver disease have higher risk of developing liver-related complications.1–3 Interferon-alfa is the first approved drug in the treatment of chronic hepatitis B. In a meta-analysis, approximately one third of patients lost HBeAg after 16 to 24 weeks of interferon treatment.4 The treatment response to interferon among Asian patients appears to be less satisfactory than that for Caucasians when patients with normal alanine aminotransferase (ALT) are included.5, 6 Nonetheless, long-term follow-up studies have shown reduced incidence of liver-related complications, hepatocellular carcinoma, and mortality among sustained interferon responders compared with nonresponders, relapsers, or untreated controls.7–11

Lamivudine and adefovir dipivoxil are 2 registered oral nucleos(t)ide analogs that can suppress replication of hepatitis B virus (HBV). The HBeAg seroconversion rate of either treatment at 1 year is approximately 15%.12–14 After cessation of lamivudine, sustained HBeAg seroconversion in Caucasians has been reported to be 72% in 3 years.15 However, in other reports, up to 50% of patients will develop virological relapse after cessation of lamivudine, which may be related to the short duration (<1 year) of lamivudine usage.16, 17 HBeAg seroconversion after adefovir dipivoxil treatment seems to be more durable (>90%) on preliminary analysis, but further data are required for confirmation.18 Extended use of nucleos(t)ide analogs may prolong the viral suppression but at the expense of development of drug resistance. Up to 70% of HBeAg-positive patients on lamivudine will develop drug resistance at 4 years.19 Although no clear long-term data exist for adefovir resistance in HBeAg-positive patients, approximately 6% of HBeAg-negative patients will have drug resistance after continuous adefovir dipivoxil treatment for 3 years.20 The incidence of drug resistance to adefovir may be higher in HBeAg-positive patients, who have higher HBV DNA levels.

In a European randomized study, combination of conventional interferon and lamivudine has no clear advantage over interferon or lamivudine treatment alone.21 In this study, interferon was given for 16 weeks, and lamivudine was given for 24 weeks only in the combination treatment group. Peginterferon alfa-2b is synthesized by covalently linked recombinant interferon alfa-2b to a straight-chain molecule of polyethylene glycol (12,000 daltons) in a 1:1 molar ratio. We have performed a randomized trial comparing peginterferon alfa-2b and lamivudine combination treatment versus lamivudine monotherapy in the treatment of HBeAg-positive chronic hepatitis B.22 On intent-to-treat analysis at 24 weeks after treatment, 36% of patients on combination treatment developed HBeAg seroconversion, which was significantly higher than that of patients on lamivudine monotherapy (14%). Whether this virological response can be sustained is uncertain. In this study, we set to investigate (1) the long-term sustained virological response after cessation of combination treatment compared with lamivudine monotherapy, and (2) the predictors of sustained virological response to combination treatment.

Abbreviations

HBeAg, hepatitis B e antigen; HBV, hepatitis B virus; ALT, alanine aminotransferase; HBsAg, hepatitis B surface antigen; PCR, polymerase chain reaction.

Patients and Methods

Patients

Ninety-six patients who had completed per protocol treatment in a randomized study of peginterferon alfa-2b (PegIntron; Schering-Plough Corp., Kenilworth, NJ) and lamivudine (Zeffix; GlaxoSmithKline, Middlesex, UK) combination versus lamivudine monotherapy were included in this study.22 The study was approved by the Human Investigation Committee of the institution, and all patients gave informed consent. At screening, all patients had treatment-naive chronic hepatitis B with positive hepatitis B e antigen (HBeAg), serum HBV DNA of at least 500,000 copies/mL and ALT 1.3 to 5 times of upper limit of normal. Patients with coinfection by hepatitis C, hepatitis D, and human immunodeficiency viruses were excluded. Peginterferon alfa-2b was prescribed as subcutaneous injection at a dose of 1.5 μg/kg/week for patients with body weight less than 65 kg or 100 μg/week for patients with body weight more than 65 kg for 32 weeks. Lamivudine was administered as 100 mg oral daily for 52 weeks in both groups of patients. In patients receiving combination therapy, peginterferon alfa-2b was administered 8 weeks before the commencement of lamivudine, then a combination of both treatments for 24 weeks followed by lamivudine monotherapy for a further 28 weeks. The presence of lamivudine-resistant mutations was determined in the serum sample at the end of treatment. We performed a liver biopsy within 4 weeks before commencement of treatment and at the end of treatment. We scored histological necro-inflammation by the Knodell scoring system (0-18) and liver fibrosis by the Ishak scoring system (0-6). We defined ALT flare during treatment as pretreatment baseline level plus ALT increase of more than 116 IU/L (i.e., 2 times upper limit of normal) at any time during treatment.23

After completion of treatment, all patients were followed up at weeks 4, 8, 16, 24, 32, 48, 52, and then every 12 weeks thereafter. Liver biochemistry was monitored at every posttreatment visit, HBeAg and antibodies to hepatitis B e antigen (anti-HBe) were monitored at every visit from week 8 onwards, and HBV DNA was monitored at weeks 4, 8, 16, 24, 52, and then every 24 weeks. Hepatitis B surface antigen (HBsAg) was monitored yearly. In this study, all patients had posttreatment follow-up for at least 52 weeks.

We defined sustained responders as patients who had persistent HBeAg loss and had less than 2 occasions with HBV DNA > 100,000 copies/mL at any time during the entire posttreatment follow-up period. We defined virological relapse as either HBV DNA > 100,000 copies at any 2 or more occasions or HBeAg reversion during the entire posttreatment follow-up period. The timing of virological relapse was taken as the timing of first HBV DNA elevation or HBeAg reversion, whichever was earlier. We defined biochemical relapse as ALT elevation to more than 2 times upper limit of laboratory normal. We defined decompensation as elevated serum bilirubin > 50 IU/L accompanied biochemical relapse.

Laboratory Assays

Serological Assays.

HBsAg was tested by commercially available ELISA kits (COBAS CORE HBsAg IIEIA, Roche Diagnostics Corp., Indianapolis, IN). Antibodies to hepatitis C, hepatitis D, and human immunodeficiency viruses were tested by commercially available ELISA (Abbott GmBH Diagnostika, Wiesbaden-Delkenheim, Germany). HBeAg and anti-HBe were measured by ELISA (Sanofi Diagnostics, Pasteur, France).

Virological Assays.

We used TaqMan real-time polymerase chain reaction (PCR) assay to measure HBV DNA levels.24, 25 The range of HBV DNA detection was from 102 to 109 copies/mL with correlation coefficient of the standard curve routinely greater than 0.990. We determined HBV genotyping by restriction fragment length polymorphism in the residual serum sample of each patient on initial visit as previously described.26, 27 We determined lamivudine-resistant mutants by the INNO-LiPA HBV DR line probe assay (LiPA) at the end of treatment according to the instruction of the manufacturer (Innogenetics N.V., Ghent, Belgium).28

Statistical Analysis

Statistical tests were performed by SPSS (version 11.0, Chicago, IL). HBV DNA was logarithmic transformed to normal distribution for analysis. Continuous variables with normal distribution were expressed as mean ± standard deviation and variables with skewed distribution as median and range. Baseline categorical variables were compared by Pearson's chi-square test, and continuous variables were compared by Student t test and Mann-Whitney U test as appropriate. We used Kaplan-Meier survival analysis to compare the probability of sustained virological response as well as HBeAg seroconversion among non-responders. We used the Cox proportional hazard model to assess the predictors of sustained virological response among patients who received peginterferon and lamivudine combination treatment. All statistical tests were two-sided. Statistical significance was taken as P < .05.

Results

Baseline Clinical Characteristics.

Forty-eight patients who received combination treatment and 48 patients who received lamivudine monotherapy were included in this study. One patient receiving lamivudine monotherapy had HBeAg seroconversion before commencement of treatment and was excluded from analysis. The baseline characteristics of patients are shown in Table 1. Patients who received combination treatment had higher ALT levels than those who received lamivudine monotherapy (P = .039). Otherwise, there was no difference in age, sex ratio, body mass index, log HBV DNA, HBV genotype distribution, and histological necroinflammatory and fibrosis scores between the 2 groups.

Table 1. Baseline Characteristics of Patients on Peginterferon and Lamivudine Combination Treatment and Lamivudine Monotherapy
 Combination TreatmentLamivudine
  1. NOTE. Forty-four patients on combination treatment and 45 patients on lamivudine monotherapy had baseline liver biopsy performed.

  2. Abbreviations: BMI, body mass index; ALT, alanine aminotransferase; HBV, hepatitis B virus.

Number of patients4847
Age32 ± 1035 ± 10
Male29 (60%)34 (72%)
BMI (kg/m2)22.7 ± 4.024.0 ± 3.8
ALT (IU/L)140 (48-1,179)118 (36-461)
Log HBV DNA (copies/mL)8.2 ± 1.07.9 ± 1.1
Genotype B:C:mixed15:30:316:28:3
Necroinflammatory score5 (1–10)5 (1–12)
Fibrosis score1 (0–6)1 (0–5)

Sustained Virological Response.

The posttreatment follow-up of patients who received combination treatment was 117 ± 34 (range, 51-172) weeks, and that of patients who received lamivudine monotherapy was 124 ± 29 weeks (range, 82-174) (P = .27). At end of treatment, 30 patients (63%) in the combination treatment group and 13 patients (28%) in the lamivudine group had lost HBeAg (P = .001) (Fig. 1). Only 1 patient who had HBeAg loss after combination treatment did not develop anti-HBe antibody; otherwise, all other patients had developed HBeAg seroconversion to positive anti-HBe. Overall, 14 patients on combination treatment and 4 patients on lamivudine monotherapy developed sustained response, and all of them had positive anti-HBe antibodies at posttreatment follow-up. The probabilities of sustained response for combination treatment and lamivudine monotherapy were 33% and 13% at week 24, 31% and 11% at week 52, and 29% and 9% at week 76, respectively (Fig. 2A). No patients developed virological relapse after week 76 in either treatment group. The cumulative probability of sustained response was significantly higher for combination treatment versus lamivudine monotherapy (log-rank test, P = .0015). Only 1 patient in the combination treatment group developed HBsAg clearance. None of the sustained responders had persistently negative HBV DNA by PCR. At the last follow-up visit, 6 (43%) of 14 patients in the combination treatment group and 2 (50%) of 4 patients in the lamivudine group had negative HBV DNA by PCR.

Figure 1.

Flow diagram of the clinical course of patients. For posttreatment biochemical relapse, decompensation, and HBeAg loss at last follow-up, patients were censored at the time of lamivudine retreatment for those who received lamivudine retreatment of hepatitis relapse. EOT, end of treatment; SVR, sustained virological response; Bch, biochemical; Decomp, decompensation; FU, follow-up.

Figure 2.

(A) Cumulative probability of overall sustained virological response. Time zero = time of cessation of treatment. Solid line indicates combination treatment (Combo); and broken line, lamivudine monotherapy (Lam) [log-rank test P = .0015]. (B) Cumulative probability of overall sustained virological response when the presence of single transient hepatitis B virus (HBV) DNA elevation was taken as non-sustained response. Time zero = time of cessation of treatment. Solid line indicates combination treatment (Combo); and broken line, lamivudine monotherapy (Lam). Log-rank test P < .00001.

Two sustained responders in the combination treatment group had transient elevation of HBV DNA to >100,000 copies/mL at weeks 8 and 16 after treatment, respectively. All 4 sustained responders in the lamivudine monotherapy group had transient elevation of HBV DNA > 100,000 copies/mL (3 patients at week 8 and 1 patient at week 100 after treatment). If all patients who had a single episode of transient HBV DNA elevation were regarded as nonresponders, the probability of sustained virological response of combination treatment group was still higher than that of lamivudine group (log rank test, P < .0001, Fig. 2B).

Four (29%) of 14 sustained responders on combination treatment and 1 (25%) of 4 sustained responders on lamivudine monotherapy had transiently elevated ALT levels, and all of them had peak ALT levels less than 2 times the upper limit of laboratory normal. At the time when ALT peaked, all 4 patients in the combination group had HBV DNA < 10,000 copies/mL, whereas the patient in the lamivudine monotherapy group had HBV DNA of 4.7 × 106 copies/mL. At the last follow-up visit, only 1 sustained responder in the combination treatment group had ALT 1.3 times the upper limit of normal, whereas all other sustained responders in either treatment group had normal ALT levels.

Among patients who have lost HBeAg at the end of treatment, the subsequent probabilities of sustained response for the combination treatment group and lamivudine monotherapy group were 53% and 46% at week 24, 50% and 38% at week 52, and 47% and 31% at week 76, respectively (log rank test, P = .36) (Fig. 3).

Figure 3.

Cumulative probability of sustained virological response among patients who have lost HBeAg at the end of treatment. Time zero = time of cessation of treatment. Solid line indicates combination treatment (Combo); and broken line, lamivudine monotherapy (Lam). Log-rank test P = .36.

Virological and Biochemical Relapse.

Thirty-four patients in the combination group and 43 patients in the lamivudine group did not develop sustained virological response (Fig. 1). All non-sustained responders had HBV > 1,000,000 copies/mL at virological relapse. Among patients who had HBeAg loss at the end of treatment, 12 (75%) of 16 patients in the combination treatment group and 8 (98%) of 9 patients in the lamivudine monotherapy group also had HBeAg reversion. Among patients who still had positive HBeAg at the end of treatment, the probabilities of spontaneous HBeAg seroconversion in the combination treatment group and the lamivudine monotherapy group were 0% and 6% at week 24, 11% and 9% at week 52, 25% and 15% at week 100, and 25% and 30% at week 148, respectively (log rank test P = .69) (Fig. 4).

Figure 4.

Cumulative probability of spontaneous HBeAg seroconversion among patients who are still HBeAg positive at the end of treatment. Time zero = time of cessation of treatment. Solid line indicates combination treatment (Combo); and broken line, lamivudine monotherapy (Lam) Log rank test P = .69.

Biochemical relapse developed in 32 (94%) of 34 non-sustained responders in the combination group and 38 (88%) of 43 non-sustained responders in the lamivudine group (P = .64) (Fig. 1). Nine patients (26%) in the combination group and 15 patients (35%) in the lamivudine group had peak ALT more than 10 times the upper limit of normal at biochemical relapse (P = .59). The peak posttreatment ALT levels were 6.1 (range, 0.7-32.3) times the upper limit of normal in the combination treatment group and 7.1 (range, 0.5-62.2) times the upper limit of normal in the lamivudine group (P = .98).

Seven patients in the combination treatment group and 14 patients in the lamivudine group received open-labeled lamivudine treatment for severe posttreatment biochemical relapse. On assessment of the virological and biochemical parameters at the last follow-up visit, patients who received retreatment of lamivudine were censored at the time of start of retreatment. At last follow-up visit, 12 (35%) of 34 non-sustained responders in the combination group and 12 (28%) of 43 non-sustained responders in the lamivudine group had negative HBeAg, 10 patients (29%) in the combination group and 9 patients (21%) in the lamivudine group had HBV DNA < 100,000 copies/mL, only 1 patient in the combination group had HBV DNA negative by PCR, and 17 (50%) of 34 patients in the combination group and 16 (37%) of 43 patients in the lamivudine group had normal ALT levels. One patient in the combination group, who required open-label lamivudine for severe posttreatment relapse and who developed acute duodenal ulcer bleeding complicated by shock and aspiration pneumonia, died at week 64 after treatment. Otherwise, no patient in either group developed ascites or hepatic encephalopathy.

Two patients who received combination treatment developed decompensation at posttreatment biochemical relapse with serum bilirubin elevated to 140 IU/L at week 32 and 138 IU/L at week 63, respectively. Four patients who received lamivudine monotherapy developed decompensation with elevated serum bilirubin from 52 IU/L to 153 IU/L at weeks 12 to 22 after treatment. Five of these 6 patients had a histological fibrosis score of 1, and the remaining patients had a fibrosis score of 4 before commencement of treatment.

Predictors of Sustained Response to Combination Treatment.

Baseline clinical and virological variables as well as peginterferon dosage and ALT flare during treatment were not associated with sustained response to combination treatment (Table 2). Among patients who had negative HBeAg at the end of combination treatment, there was a trend for longer duration of HBeAg seroconversion before cessation of treatment to associate with sustained response (Table 3). Otherwise, the end-of-treatment ALT level, HBV DNA load, and histology had no association with sustained response.

Table 2. Baseline Characteristics of Patients on Peginterferon and Lamivudine Combination Treatment With Respect to Sustained Virological Response
 Sustained ResponderNon-sustained ResponderP Value
  1. NOTE. Thirteen sustained responders and 31 non-sustained responders had baseline liver biopsy performed. Two sustained responders and 1 non-sustained responder has mixed genotype B and C HBV.

  2. Abbreviations: BMI, body mass index; ALT, alanine aminotransferase; HBV, hepatitis B virus.

Number of patients1434 
Age30 ± 932 ± 10.087
Male7 (50%)24 (67%).46
BMI (kg/m2)23.0 ± 4.822.6 ± 3.7.77
ALT (IU/L)182 (57-420)134.5 (48-1179).92
Log HBV DNA (copies/mL)7.9 ± 0.78.4 ± 1.1.57
Genotype B:C4:811:22.47
Necroinflammatory score7 (1–10)5 (1–10).40
Fibrosis score1 (1–4)1 (0–6).59
Peginterferon dose (μg/kg)89 ± 1691 ± 9.42
ALT flare during treatment3 (21%)9 (27%).71
Table 3. Clinical Characteristics of Patients Who Had Negative HBeAg at End of Peginterferon and Lamivudine Combination Treatment With Respect to Sustained Virological Response
 Sustained ResponderNon-sustained ResponderP Value
  1. NOTE. All 14 sustained responders and 14 of 16 non-sustained responders had liver biopsy at end of treatment.

  2. Abbreviations: BMI, body mass index; ALT, alanine aminotransferase; HBV, hepatitis B virus; HBeAg, hepatitis B e antigen.

Number of patients1416 
ALT (IU/L)26.5 (14-65)30 (18–65)1.00
Log HBV DNA (copies/mL)3.4 ± 1.23.9 ± 0.7.49
rtM204I/V4 (29%)3 (19%).57
Necroinflammatory score2 (1–5)2 (0–6).34
Fibrosis score1 (0–6)1 (0–5).87
Weeks of HBeAg seroconversion before stopping treatment39 ± 725 ± 20.090

Discussion

Use of peginterferon either as monotherapy or in combination with lamivudine has good clinical efficacy in the treatment of chronic hepatitis B.22, 29–31 The reported rates of HBeAg loss at 24 weeks after treatment ranged from 27% to 36%, which was significantly higher than that achieved by lamivudine monotherapy (range, 14%-19%).22, 31 Addition of lamivudine to peginterferon treatment can improve the on-treatment viral suppression, but the advantage is not sustained after treatment cessation.29, 30 From previous experience with conventional interferon, virological relapse is uncommon after cessation of interferon treatment.32 On the contrary, the HBeAg loss after lamivudine monotherapy is not sustained, and virological relapse continues to occur up to 3 years after treatment.16, 33 No long-term data have been reported on the sustained virological response after peginterferon and lamivudine combination treatment. In this study with up to 3-year posttreatment follow-up, we have confirmed higher sustained virological response among patients who received peginterferon and lamivudine combination treatment (29%) compared with those who received lamivudine monotherapy (9%).

Although HBeAg loss has been used as an indicator of treatment response, the sustained response after stopping treatment is more important. In this study, most virological relapses occurred within the initial 24 weeks after cessation of peginterferon and lamivudine combination treatment. This high rate of virological relapse is not typical of interferon treatment but may be related to the use of lamivudine. Only a minority of patients who remained HBeAg-negative at week 24 after treatment developed virological relapse up to week 76, and no patient had virological relapse thereafter. On the contrary, HBeAg loss after cessation of lamivudine monotherapy was not sustained, and virological relapse continued to occur up to the second year after treatment.

In this study, no predictor could be identified for sustained response to combination peginterferon and lamivudine treatment. This is in contrast to the findings of another multi-center trial on peginterferon alfa-2b in chronic hepatitis B in which HBV genotype, low HBV DNA, and high ALT level were found to be independent predictors of HBeAg loss 24 weeks after cessation of treatment.29 Possible explanations for the difference in the findings of these 2 trials include our small sample size, difference in definition of treatment response, duration of follow-up, patient ethnic background, and difference in distribution of genotype B and genotype C HBV subgroups.34, 35

We defined sustained response as sustained HBeAg loss and HBV DNA less than 100,000 copies/mL.36 These patients should have milder histological damage as compared with patients with higher HBV DNA levels.37 However, HBeAg reversion and biochemical relapses have been reported with HBV DNA lower than 100,000 copies/mL, and a lower HBV DNA level, probably less than 10,000 copies/mL, has been suggested to indicate inactive liver disease.38, 39 In this study, all sustained responders have several readings of HBV DNA available for at least 1 year after treatment. Nonetheless, a minority of them did have transient elevation of HBV DNA and mild biochemical activity, and longer follow-up may be necessary to clearly delineate their risk of disease reactivation and clinical outcome. We believe that HBeAg loss secondary to peginterferon is sustained and associated with good long-term prognosis, as in the case of conventional interferon.7, 8

Based on the results of our study, we recommend that the peginterferon and lamivudine combination can be stopped at the end of 1 year if HBeAg loss is achieved. Close observation is necessary, particularly in the initial 24 weeks, when disease reactivation occurs most frequently. Because severe disease reactivation can occur among patients who are still HBeAg positive at the end of 1 year of treatment, further continuation of lamivudine can be considered, but the emergence of drug resistance will become a concern. More studies are required to investigate the benefit of extended lamivudine treatment, because 25% of non-responders at the end of treatment will develop spontaneous HBeAg loss in the subsequent 3 years.

Our study has several limitations. First, because the posttreatment follow-up is only up to 3 years, we cannot compare the clinical outcomes including hepatocellular carcinoma and mortality among sustained responders versus relapsers or non-responders. Although we believe sustained responders should have better prognosis, much longer follow-up will be required to address this important question. Second, a staggered regimen of 32-week peginterferon and 52-week lamivudine combination was used in our study. The sustained response of a longer duration of peginterferon treatment or lamivudine treatment is uncertain. Whether the sustained response will be different if the 2 drugs are started simultaneously will also await data from other studies.29, 31 Third, our data cannot be directly extrapolated to that of HBeAg-negative patients. The posttreatment relapse rate of HBeAg-negative patients to conventional interferon treatment is much higher and may occur up to the second year,40 and we also anticipate similar phenomenon to occur in peginterferon and lamivudine combination treatment.

In conclusion, 32% of patients received combination peginterferon alfa-2b and lamivudine treatment; they can achieve sustained HBeAg loss and HBV DNA suppression up to 3 years after treatment. This is significantly higher than the sustained response to lamivudine monotherapy. Most virological relapse occurs within 24 weeks of stopping combination treatment. Further studies with longer follow-up will be needed to investigate the impact of this combination regimen on clinical outcomes of chronic hepatitis B.

Ancillary