The relative clinical advantage of stavudine therapy over zidovudine therapy has not been identified. We sought to determine whether the initiation of stavudine therapy (in patients who had previously received 6 or more months of zidovudine therapy) compared with continued treatment with zidovudine would reduce the subsequent risk for clinical progression.

Methods

Top Methods Results Discussion Author & Article Info References

Study Design

In this multicenter, randomized, double-blind trial, we compared stavudine with continued zidovudine therapy in HIV-infected patients who had previously received zidovudine. Randomization was done centrally by using a computer randomization program; stratification was done according to study site and three baseline CD4⁺ cell strata: 100 cells/mm³ or fewer, 101 to 300 cells/mm³, and more than 300 cells/mm³.

Once randomization was accomplished, the corresponding patient identification number was transmitted to the investigative site by telephone or fax. Patients were assigned treatment kits through a central drug allocation system that was accessed by telephone. This system matched the patient identification number with the blinded assigned treatment and identified an appropriate blinded treatment kit within the inventory of the investigative site. The same system was used to obtain a new blinded kit assignment for continued therapy.

Patients

Patients with HIV infection were eligible for the study if they were 13 years of age or older and had CD4⁺ cell counts of 50 to 500 cells/mm³. Patients were required to have received zidovudine for at least 6 months and to have tolerated a dosage of at least 500 mg/d.

Patients had to be in stable clinical condition without acute acquired immunodeficiency syndrome (AIDS)-defining opportunistic infection; intractable diarrhea; or the need for ongoing therapy with myelosuppressive, neurotoxic, cytotoxic, or hepatotoxic agents. Patients were excluded if they had previously received therapy with didanosine, zalcitabine, or stavudine or if they had a history of bilateral peripheral neuropathy. Women of childbearing potential were required to have negative results on serum pregnancy tests, and men and women were required to use effective contraception. Pregnant or breastfeeding women were excluded.

Before enrollment, all patients signed a document of informed consent that had been approved by an institutional review board.

Study Medication

Stavudine capsules were administered orally in two daily doses of 40 mg to patients weighing 60 kg or more and 30 mg to patients weighing less than 60 kg. This corresponded to a total dosage of approximately 1.0 mg/kg of body weight per day. Patients who were randomly assigned to zidovudine received capsules that were indistinguishable in appearance from the stavudine capsules but that contained only inert filler. Zidovudine capsules were administered in three daily doses of 200 mg (600 mg/d). Patients who were randomly assigned to stavudine received capsules that were indistinguishable in appearance from the zidovudine capsules but that contained only inert filler.

No treatment crossover occurred. If the CD4⁺ cell count declined by 50% in any patient, the patient could elect to stop therapy with the study medication. The study medication was also stopped if a patient developed a new AIDS-defining event. Patients who stopped therapy were allowed to take any alternative antiretroviral treatment, including experimental therapy. Didanosine was provided at no cost when it was the preferred treatment option. Prophylaxis against Pneumocystis carinii pneumonia was given in accordance with guidelines from the Centers for Disease Control and Prevention [9]. Prophylaxis against other opportunistic infections, including Mycobacterium avium complex infection, toxoplasmosis, or tuberculosis, was allowed at the discretion of the patient's physician.

Study Procedures

Eligibility screening and procedures done at baseline included three determinations of CD4⁺ cell count (one for eligibility and two to establish a baseline value); chest radiography; standard hematologic and serum chemistry panels; measurements of serum levels of ß₂-µglobulin, immune complex-dissociated p24 antigen, and neopterin; and physical examination. Patients were seen 2 and 4 weeks after they began receiving the study medication and every 4 weeks thereafter. Follow-up visits included a brief physical examination, relevant history, and repetition of some laboratory tests. If patients discontinued treatment with the study medication for any reason, monthly visits continued until the end of the trial and body weight, CD4⁺ cell count, use of antiretroviral therapy or prophylactic medication, and the development of any HIV-related events were assessed.

Study End Points

The clinical end point of the trial, clinical progression, was defined as all occurrences of AIDS-defining events (as determined by the 1987 definition from the Centers for Disease Control and Prevention [10]) or death. Time to first AIDS-defining event or death was also determined to facilitate comparison with other trials. Additional efficacy end points included death and an aggregate end point (treatment failure) that was defined as the first occurrence of an AIDS-defining event, death, or a confirmed decline in baseline CD4⁺ cell count of 50% or more [11]. Changes in weight, CD4⁺ cell count, and p24 antigen level were also evaluated. All end points were reviewed by a medical monitor who was blinded to treatment assignment.

Data Safety and Monitoring Board

An interim review of safety and efficacy was done by an independent data safety and monitoring board using predefined stopping rules that were based on the O'Brien and Fleming implementation of the Lan-Demets procedure [12]. Accumulating data were reviewed for safety in March 1993 and were reviewed for safety and efficacy in July 1993 and February 1994. On the basis of the rate of clinical events (AIDS-defining events and deaths) in the study, the data safety and monitoring board recommended that the study last 32 months (until December 1994), so that a conclusion could be reached on the basis of clinical progression. Interim results and deliberations of the data safety and monitoring board were not available to anyone involved in the conduct of the trial. A preliminary analysis that was limited to surrogate markers and safety (and used data from 6 months of follow-up of the first 359 patients enrolled in the trial) was presented to the Food and Drug Administration in June 1994 under the accelerated approval process.

Industry Role

Employees of Bristol-Myers Squibb participated as co-investigators in the design, analysis, and presentation of these data. These employees participated in the writing of this paper; the company did not have the right to approve or disapprove publication of the completed manuscript.

Statistical Analysis

Statistical analyses were done using SAS (SAS Institute, Cary, North Carolina) and S-Plus (Statistical Sciences Inc., Seattle, Washington) software. All patients who received a treatment assignment were included in the efficacy analyses according to the intention-to-treat principle. Safety analyses were restricted to patients who received a study drug (stavudine or zidovudine). The distribution of baseline characteristics was examined using the Fisher exact test for categorical data and the Wilcoxon test for continuous data [13, 14].

Longitudinal data for CD4⁺ cell counts were analyzed to 96 weeks, after which censoring exceeded 50%. Two nonparametric tests were used: the generalized Wilcoxon test, which tests for the equality of two distributions of measurements, and the stochastic ordering test, which tests for a consistent difference in the distribution of measurements [15].

The product-limit method of Kaplan and Meier [16] was used to estimate the time-to-event distributions for study end points, adverse events, and duration of therapy with the study drug. Time-to-event analyses were stratified to incorporate the baseline CD4⁺ cell stratification that was used for randomization ( 100 cells/mm³, 101 to 300 cells/mm³, and >300 cells/mm³). Diagnosis of AIDS at baseline was included as a second stratification variable for time-to-study end points on the basis of an investigation of the proportionality assumptions in these models [17]. For analysis of multiple events, P values, relative risks, and 95% CIs were based on the maximum likelihood estimates that were determined from the generalized Cox proportional-hazards model as developed by Wei and colleagues [18]. This model allows for the analysis of several events by combining the results of four marginal Cox models: times to first, second, and third AIDS events and time to death. Overall estimates of relative risk and CIs, which consider the correlation of events, were computed under the assumption of a common treatment effect.

The estimates of relative risk and CIs for time to first event were based on the maximum likelihood estimates from the Cox proportional-hazards model, and P values for treatment comparisons were determined by using stratified log-rank tests [16]. Additional models, which adjusted for prognostic baseline covariates, were constructed.

Two-sided P values are reported. A nominal significance level (P = 0.0463) for the analyses of clinical progression preserves the experiment-wise type I error rate of 5% after the two interim analyses.

Results

Top Methods Results Discussion Author & Article Info References

Demographic and Clinical Characteristics of Patients

Between May 1992 and January 1993, 822 patients were enrolled in the trial. The last day of patient follow-up was 31 December 1994. Eight patients never started blinded treatment but are included in the efficacy analysis. No clinically significant differences were seen in demographic or clinical characteristics between the treatment groups (Table 1). The frequency and type of medication taken for prophylaxis against opportunistic infections were similar at study entry and at the end of the trial in the two treatment groups (Table 2). The median duration of previous zidovudine therapy was 88 weeks.

Patients were monitored for clinical progression and survival for the duration of the trial, regardless of whether they reached a trial end point or discontinued therapy with the study drug. Ninety-two percent of patients were followed until the end of the trial: Six hundred twenty-five (76%) had a final study visit between October and December 1994, and 128 (16%) died before the end of the trial. The remaining 69 patients (8%) were lost to follow-up before the end of the study. The median duration of follow-up was 115 weeks for both treatment groups. Patient experience with study medication is shown in Table 1. Among patients who started treatment, the median duration of therapy with the study drug was 79 weeks for the stavudine group and 53 weeks for the zidovudine group (P < 0.001).

Progression to Study End Points

The rate of clinical progression to one or more events (AIDS-defining event or death) was lower in the stavudine group than in the zidovudine group (relative risk, 0.75 [CI, 0.58 to 0.98]; P = 0.03) (Table 3 and Figure 1). Two hundred thirty-one events occurred among 417 patients in the stavudine group (26 events per 100 person-years), and 262 events occurred among 405 patients in the zidovudine group (32 events per 100 person-years). In other words, 1 clinical event was prevented for every 17 patients who were switched from zidovudine therapy to stavudine therapy. The most common clinical events were death (146 patients), invasive candidiasis (66 patients), P. carinii pneumonia (63 patients), cytomegalovirus infection (62 events), disseminated Mycobacterium avium complex infection (37 patients), and the wasting syndrome (25 patients).

View larger version (21K): [in this window] [in a new window]   Figure 1. Kaplan-Meier estimate of the distribution of time to clinical progression (all acquired immunodeficiency syndrome-defining events or death) by treatment group. Top. Results for all study patients. Bottom. Results stratified by the three protocol-defined strata of CD4+ cell count at baseline ( 100 cells/mm3, 101 to 300 cells/mm3, and >300 cells/mm3).

When the analysis of time to clinical progression was restricted to time to first AIDS-defining event or death, the reduction in risk for progression in the stavudine group was of similar magnitude: 16 events per 100 person-years in the stavudine group compared with 18 events per 100 person-years in the zidovudine group (relative risk, 0.78 [CI, 0.60 to 0.998]; P = 0.04).

A trend toward longer survival among patients receiving stavudine was seen. Sixty-seven of 417 patients (8 patients per 100 person-years) in the stavudine group died compared with 79 of 405 patients (10 patients per 100 person-years) in the zidovudine group (relative risk, 0.74 [CI, 0.53 to 1.02]; P = 0.066) (Table 3).

Treatment failure ( 50% decline in CD4⁺ cell count or occurrence of a first AIDS-defining event or death) was less frequent among patients receiving stavudine than among those receiving zidovudine (relative risk, 0.68 [CI, 0.56 to 0.82]; P < 0.001) (Table 3 and Figure 2). Forty-two percent of the treatment failure events were a decline in CD4⁺ cell count of 50% or greater.

View larger version (22K): [in this window] [in a new window]   Figure 2. Kaplan-Meier estimate of the distribution of time to treatment failure (first acquired immunodeficiency syndrome-defining event, death, or sustained decline in CD4+ cell count to 50% of baseline level) by treatment group. Top. Results for all study patients. Bottom. Results stratified by the three protocol-defined strata of CD4+ cell count at baseline ( 100 cells/mm3, 101 to 300 cells/mm3, and >300 cells/mm3).

Differences between treatment groups that favored the stavudine group in progression to study end points were seen among patients in each of the three protocol-defined CD4⁺ cell strata ( 100 cells/mm³, 101 to 300 cells/mm³, and >300 cells/mm³) and among patients in various clinical stages of HIV disease (asymptomatic disease, symptomatic disease, and AIDS) (Table 3, Figure 1 and Figure 2). Differences in progression were most apparent among patients who had CD4⁺ cell counts of 100 cells/mm³ or less or greater than 300 cells/mm³ at baseline or patients who had a diagnosis of AIDS at baseline. The benefit of switching to stavudine was not significantly correlated with the duration of previous zidovudine therapy.

At 16 weeks, the weight of patients receiving stavudine had increased by a mean of 1 kg, and the weight of patients receiving zidovudine had decreased by a mean of 0.4 kg. This difference was maintained throughout the study period (P < 0.001 by stochastic ordering for the 96 weeks).

Surrogate Markers

Longitudinal plots of the mean change in CD4⁺ cell counts show a clear difference between treatment groups (Figure 3). Stavudine therapy was associated with an increase in mean CD4⁺ cell count at week 2 that was 21 cells/mm³ above baseline; the CD4⁺ cell count was sustained above baseline levels for approximately 16 weeks. In contrast, zidovudine treatment was associated with a steady decline in CD4⁺ cell counts from their baseline levels, reaching a difference of 30 to 50 cells/mm³ between treatment groups that was maintained for 96 weeks of follow-up (P < 0.001). Similar differences in CD4⁺ cell counts between treatment groups were seen in each of the patient subgroups defined by baseline CD4⁺ cell count, clinical severity of HIV disease, and sex (data not shown).

View larger version (16K): [in this window] [in a new window]   Figure 3. Mean change in CD4+ cell count from baseline values by treatment group. Bars represent SEs.

Median levels of p24 antigen among patients with immune complex-dissociated p24 antigenemia at baseline were similar in the stavudine group (119 pg/mL) and the zidovudine group (155 pg/mL). A median decrease in serum p24 antigen levels of 10 pg/mL (8% below baseline value) was seen in the stavudine group at week 4; a median change of 0 pg/mL was seen in the zidovudine group (P = 0.06). However, the p24 antigen level in the stavudine group eventually returned to the baseline value; the difference between the two treatment groups was not significant by the generalized Wilcoxon test or stochastic ordering test for 56 weeks of follow-up (P > 0.2).

Adverse Events

Anemia and neutropenia were more common among zidovudine recipients than among stavudine recipients (Table 4). Zidovudine recipients were more likely than stavudine recipients to report nausea and vomiting (178 compared with 160 patients; 35 compared with 26 complaints per 100 person-years; P < 0.01) and myalgia (139 compared with 132 patients; 27 compared with 21 complaints per 100 person-years; P > 0.02). Stavudine recipients were more likely than zidovudine recipients to report flu-like symptoms (62 compared with 32 patients; 10 compared with 6 complaints per 100 person-years; P > 0.02) and peripheral neuropathy (56 compared with 18 patients; 9 compared with 4 complaints per 100 person-years; P < 0.001).

Fifty-six patients receiving stavudine (12%) and 18 receiving zidovudine (4%) developed peripheral neuropathy that required a modification of dose during the course of treatment (Table 4). Neuropathy was two to four times more common among patients who had a diagnosis of AIDS at baseline than among patients who were asymptomatic at baseline. Among patients with AIDS, the 1-year rates of neuropathy were 25% for the stavudine group and 12% for the zidovudine group; among asymptomatic patients, the 1-year rates were 10% for the stavudine group and 3% for the zidovudine group. Of the 56 patients receiving stavudine who developed neuropathy, 35 (63%) had complete resolution of neurologic signs and symptoms within a median of 17 days (range, 4 to 57 days). These patients resumed stavudine therapy, usually at a reduced dose, and tolerated treatment for a median of 23 additional weeks (range, 1 to 120 weeks).

Discussion

Top Methods Results Discussion Author & Article Info References

This trial shows that treating HIV-infected adults with stavudine compared with continuing zidovudine therapy significantly decreases the rate of clinical progression. The risk for death was reduced by 26% among stavudine recipients (P = 0.066). The effects of stavudine could be seen across all patient strata defined by CD4⁺ cell counts at baseline and stage of HIV disease. Neuropathy was the only notable side effect that was more frequently associated with stavudine than with zidovudine, but most patients who developed neuropathy had complete resolution of symptoms after interruption of treatment and tolerated further therapy at a reduced dose.

Previous trials have shown that switching from zidovudine to didanosine was superior to continuing zidovudine therapy in patients who had previously received zidovudine and were clinically stable after 8 to 16 [19] or 16 or more [20] weeks of zidovudine therapy. Switching from zidovudine to didanosine was also shown to be superior among patients who had previously received zidovudine and were clinically deteriorating after at least 6 months of zidovudine therapy [21]. In these trials, no clear benefit of switching to didanosine could be shown in the subgroups with the most advanced disease. In contrast, we saw a clear advantage of switching to stavudine among patients with CD4⁺ cell counts of 50 to 100 cells/mm³ and among those with AIDS at study entry. Recent experience with drug combinations that include protease inhibitors conclusively shows that patients with advanced disease can benefit from antiretroviral therapy [22].

The benefit of switching to stavudine was least apparent among patients in the subgroup with 101 to 300 CD4⁺ cells/mm³. One possible reason may have been that patients in this group were more likely to receive aggressive antiretroviral therapy after treatment with the study medication was discontinued than were patients with more advanced disease at entry. This explanation is supported by several exploratory analyses. The rates of clinical progression and progression to protocol-defined treatment failure among patients with 101 to 300 CD4⁺ cells/mm³ at entry were significantly lower among the stavudine recipients for the first 64 weeks (Figure 1 and Figure 2). After this time, the hazard rate of progression decreased among patients in the zidovudine group. Almost 60% of patients in the zidovudine group were no longer receiving zidovudine by week 64, and they were more likely than patients in the stavudine group to have received combination antiretroviral chemotherapy after the study drug. If the analysis of clinical progression was censored 30 days after therapy was stopped (on-drug analysis), the treatment effect of stavudine among patients with 101 to 300 CD4⁺ cells/mm³ was greater (data not shown).

We believe that the differences seen between treatment groups across various clinical and laboratory measures were partly related to the study design, follow-up, and method of analysis. An intensive effort to collect protocol-specified data when patients were and were not receiving study medication resulted in complete follow-up to the end of the study for 92% of the participants. In addition, our primary efficacy analysis allowed for the consideration of all clinical end point events rather than just the first event. This analysis assumes that the number of subsequent AIDS-defining events over time is an important indicator of the longitudinal course of the disease. If the analysis is restricted to the time to first event, as has been done in previous trials, the relative risk of the two drugs is very similar, but the decreased number of end points results in a less precise estimate of treatment effect. Of note is that the rate of progression to the first AIDS-defining event or death among zidovudine recipients in our trial was substantially lower than the rate among CD4⁺ cell count-matched zidovudine recipients in ACTG (AIDS Clinical Trials Group) 116B/117 [20] and Bristol-Myers Squibb-010 [21], which were done 3 years earlier. This may reflect the effect of better prophylaxis against opportunistic infections. As the ability to prevent opportunistic infections continues to improve, antiretroviral therapy improves, and the frequency of opportunistic events declines, larger and longer clinical trials become necessary to capture enough end points to assess clinical effects. Accordingly, refinement of protocols and analytic methods and the use of virological markers will be critical aspects of clinical trials in the future.

In conclusion, our trial shows the relative safety and efficacy of changing from zidovudine to stavudine monotherapy in comparison with continuing zidovudine monotherapy. Although recent data clearly show that combinations of zidovudine with other antiretroviral agents are superior to zidovudine monotherapy [11, 23], many patients will not or cannot take zidovudine. For these patients, combined therapy with stavudine is a reasonable choice. Stavudine is an important agent to be considered for inclusion in combination therapy because of its low frequency of adverse reactions, convenient dosing (twice daily with or without food), penetration into the cerebrospinal fluid [24], and limited interaction with other drugs. Combination therapy with stavudine and didanosine has shown marked antiviral activity in a preliminary trial [25]. Combinations of stavudine and protease inhibitors, stavudine and lamivudine, and stavudine and zidovudine are being evaluated in clinical trials. The challenge will be to incorporate various antiretroviral agents into practical combination regimens that provide lasting efficacy against HIV disease and have minimal toxicity.

Appendix

The Bristol-Myers Squibb Stavudine/019 Study Group includes the following principal investigators: B. Akil, Southwest Combat Group, Los Angeles, California; D. Allan, New England Deaconess Hospital, Boston, Massachusetts; G. Beall, Harbor UCLA Medical Center, Torrance, California; A. Berry, University of Texas, San Antonio, Texas; R. Cantey, Medical University of South Carolina, Charleston, South Carolina; M. Dube, Los Angeles County-USC Medical Center, Los Angeles, California; E. Daar, Cedars-Sinai Medical Center, Los Angeles, California; K. Fife, Indiana University School of Medicine, Indianapolis, Indiana; I. Frank, University of Pennsylvania, Philadelphia, Pennsylvania; J. Goldsmith, Children's Hospital of Los Angeles, Los Angeles, California; G. McKinley, St. Lukes-Roosevelt Hospital Center, New York, New York; D. McMahon, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania; C. Alexander, Chase-Brexton Clinic, Baltimore, Maryland; P. Hawley, Whitman-Walker Clinic, Washington, D.C.; J. Jacobson, Department of Veterans Affairs Medical Center, Bronx, New York; D. Busch, California Pacific Medical Center, San Francisco, California; R. Lawrence, University of California Davis Medical Center, Sacramento, California; S. Marlowe, West Paces Clinical Research, Inc., Atlanta, Georgia; G. Mertz, University of New Mexico, Albuquerque, New Mexico; D. Mildvan, Beth Israel Hospital, New York, New York; R. Murphy, Northeastern University, Chicago, Illinois; D. Peterson, University of Texas Southwestern Medical Center, Dallas, Texas; J. Pottage, Rush Presbyterian-St. Lukes Hospital, Chicago, Illinois; H. Sacks, Mount Sinai Medical Center, New York, New York; D. Kuritzkes, University of Colorado, Denver, Colorado; G. Simon, George Washington University, Washington, D.C.; K. Squires, Cornell University Medical Center, New York, New York; R. Steigbigel, State University of New York, Stony Brook, New York; D. Sweet, University of Kansas, Wichita, Kansas; S. Swindells, University of Nebraska Medical Center, Omaha, Nebraska; J. Bartlett, Duke University Medical Center, Durham, North Carolina; B. Yangco, Infectious Disease Research Institute, Tampa, Florida; B. Bernstein, Milwaukee County Medical Complex, Milwaukee, Wisconsin; E. Goldstein, University of Kansas School of Medicine, Kansas City, Kansas; A. Stein, Community Research Initiative of South Florida, South Miami, Florida; J. Stern, Pennsylvania Hospital, Philadelphia, Pennsylvania; M. Fischl, University of Miami School of Medicine, Miami, Florida; C. Katlama, Hopital Pitie-Salpetriere, Paris, France; W. Rozenbaum, Hopital Rothschild, Paris, France; J. Modai, Hopital Saint-Louis, Paris, France; S. Pauluzzi, Universita di Perugia, Perugia, Italy; F.M. Gritti, Ospedale Maggiore, Bologna, Italy; F. Mazzotta, Ospedale di Careggi, Firenze, Italy; G. Scalise, Universita di Ancona, Ancona, Italy; A. Lazzarin, Ospedale S. Raffaele, Milano, Italy; D. Bassetti, Universita di Genova, Genova, Italy; M. Moron, Universita di Milano, Milano, Italy; R. Ciammarughi, Ospedale degli Infermi, Milano, Italy; A. LaMarca, Therafirst Medical Center, Fort Lauderdale, Florida; C. Rameiriz-Ronda, University of Puerto Rico and the San Juan Veterans Administration Hospital, San Juan, Puerto Rico; and M. Messina and L. Smaldone, Bristol-Myers Squibb Pharmaceutical Research Institute, Wallingford, Connecticut.

From the University of Utah Health Sciences AIDS Center, Salt Lake City, Utah; University of Pittsburgh School of Medicine and the Veterans Affairs Medical Center, Pittsburgh, Pennsylvania; Northwestern University School of Medicine, Chicago, Illinois; Houston Clinical Research Network and Park Plaza Hospital, Houston, Texas; Nalle Clinic, Charlotte, North Carolina; Hopital de L'Archet, Nice, France; and the Bristol-Myers Squibb Pharmaceutical Research Institute, Wallingford, Connecticut.

Presented in part at the eighth meeting of the International Society for Antiviral Research, Sante Fe, New Mexico, 23-28 April 1995; and the 35th Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, California, 17-20 September 1995.

Dr. Mellors: University of Pittsburgh Medical Center, Montefiore, Montefiore Infectious Diseases, W-638, 200 Lothrop Street, Pittsburgh, PA 15213-2582.

Dr. Murphy: 303 East Superior Street, Suite 828, Chicago, IL 60611.

Dr. Gathe: Houston Clinical Research Network, 215 Westheimer Street, Houston, TX 77006.

Dr. Stool: 1200 Binz, Suite 1260, Houston, TX 77004.

Dr. Jemsek: Nalle Clinic, 1918 Randolph Road, Charlotte, NC 28204.

Dr. Dellamonica: Hopital de L'Archet, Nice, France.

Drs. Cross and Dunkle: Bristol-Myers Squibb, 5 Research Parkway, Wallingford, CT 06492.