The Effect of High-Dose Saquinavir on Viral Load and CD4+ T-Cell Counts in HIV-Infected Patients

Methods

Persons who were positive for HIV type 1, were 18 years of age or older, had CD4+ T-cell counts of 200 to 500 cells/mm³, and had no active opportunistic infections were eligible for the study. Saquinavir was dispensed to patients as 200-mg capsules twice weekly and then once monthly. Compliance was monitored by patient report and capsule count.

Toxicity

Patients were initially monitored three times a week, then twice a week, and then once a month for any reported symptoms or signs of drug toxicity. Frequent laboratory testing was also done; tests included measurement of complete blood and platelet counts; serum chemistry tests; liver function tests; tests for amylase, triglyceride, and creatinine phosphokinase levels; and urinalysis. In patients with grade 3 toxicity, drug therapy was briefly discontinued and then restarted.

Virology

Quantitative peripheral blood mononuclear cell cultures were done by incubating serial fivefold dilutions of patient peripheral blood mononuclear cells (starting with 1 × 10⁶ cells) in duplicate with 1 × 10⁶ phytohemagglutin-stimulated normal peripheral blood mononuclear cells for 14 days. This was done according to the AIDS (acquired immunodeficiency syndrome) Clinical Trials Group consensus protocol for quantitative microcultures [24]. Measurements of p24 antigen levels were made for each dilution by using a commercially available p24 antigen kit (Abbott Diagnostics, Chicago, Illinois), and the results were expressed as infectious units per million peripheral blood mononuclear cells.

Plasma HIV RNA levels were measured by using a previously described reverse transcriptase polymerase chain reaction (PCR) technique [25] that has been validated in a multicenter study [26]. Duplicate plasma samples were subjected to ultracentrifugation, and the pellets were extracted by using phenolchloroform. The resulting RNA pellets were reverse transcribed along with a standard curve of known RNA copy number and then amplified by PCR with gag-specific primers. The amount of product in each reaction was measured using a nonisotopic enzyme hybridization assay and was expressed as optical density. The standard curve was generated by plotting the number of RNA copies against the optical density, and the Equation describing the curve was used to calculate the numbers of RNA copies in the patient samples. These numbers were expressed as log RNA copies/mL of plasma.

Serum p24 antigen levels were measured by Immunodiagnostic Laboratories (Hayward, California) using an enzyme-linked immunoassay system with an immune-complex disassociation step.

Peripheral blood mononuclear cell viral DNA levels were measured using a previously described quantitative PCR technique [27]. Aliquots of 1 × 10 (⁶) peripheral blood mononuclear cell pellets were lysed with proteinase K, and 250 000 cell equivalents were amplified in duplicate with a standard curve of known DNA copy number. The amount of product in each reaction was measured using a nonisotopic enzyme hybridization assay and expressed as optical density. The standard curve was generated by plotting the number of DNA copies against optical density, and the Equation describingthe curve was used to calculate the number of DNA copies in the patient samples. These numbers were corrected for percentages of cells that are CD4 cells and expressed as log DNA copies per million CD4 cells.

Immunology

CD4+ T-cell counts were measured by the AIDS Clinical Trials Group-qualified flow cytometry laboratory at Stanford University Hospital. A screening measurement and two baseline measurements (obtained 2 weeks apart) were done. The average of these three results was used as the baseline value. CD4+ T-cell counts were obtained monthly at the same time that blood was drawn for virologic tests.

Mutations

The presence of mutations at codon 48 (G→V) and 90 (L→M) in the plasma of patients was determined by using a selective PCR method similar to that used for the reverse transcriptase gene [28, 29]. Cryopreserved plasma was extracted as previously described [25] and was reverse transcribed using primer TGGAGTATTGTATGGATTTTCAG (Pro1). The complementary DNA was then amplified by using PCR under standard conditions with primer CAGAGCCAACAGCCCCACCA (Pro2). Five µL of the 582-base pair first-round PCR product was then amplified with primers specific for the wild-type and mutant sequences at each codon. For codon 48, primers CTTCCTTTTCCATCTCTGTA (IR48) and TGGAAACCAAAAATGACAGG (48WT) were used to determine whether a wild-type sequence was present, and IR48 and TGGAAACCAAAAATGACAGT (48MU) were used to determine whether a mutant sequence was present. For codon 90, primers GAAGCTCTATTAGATACAGG (IR90) and GTGCAACCAATCTGAGCCAA (90WT) were used to determine whether a wild-type sequence was present, and IR90 and GTGCAACCAATCTGAGCCAT (90MU) were used to determine whether a mutant sequence was present. Twenty µL of the PCR product from each of the second set of PCR reactions was analyzed for genotype on 3.0% agarose gel with ethidium bromide staining. The PCR products were determined to have a mutant or wild-type sequence according to the method described by Boucher and colleagues [30, 31] and Larder and associates [28]. A sample was considered to contain the codon 48 wild-type sequence if amplification with primers IR48 and 48WT resulted in a 309-base pair product. A sample was considered to contain the codon 90 wild-type sequence if amplification with primers IR90 and 90WT resulted in a 228-base pair product. A sample was considered to contain the codon 90 mutant sequence if amplification with primers IR90 and 90MU resulted in a 228-base pair product. Samples showing bands in both the wild-type and mutant reactions were re-evaluated using 5 µL of a 1:20 and 1:400 dilution of the first-round PCR product in a second-round reaction. Samples showing only a wild-type or mutant product in the dilution reactions were scored as wild-type or mutant, respectively. Samples showing both the wild-type and mutant products in the dilution reactions were considered to be mixtures. Samples in which a mixture was detected were reported as mutant for the purposes of the analysis. Patients showing a mutation at either codon 48 or codon 90 at week 24 were assayed at earlier time points to determine the timing of the appearance of the mutation.

The selective PCR assays were validated by complete sequencing of the protease gene on a subset of 38 samples by using standard dye-terminator sequencing. The first-round PCR product used in the selective PCR assay was used to generate a second PCR fragment using primers for -PRO4 (TGTAAAACGACGGCCAGTAGCCCCACCAGAAGAGAGCTT), which contained the M13(− 20) primer sequence, and primer IR. This product was sequenced by using the M13 forward primer and dye-labeled dideoxy terminators with an Applied Biosystems Model 370A (Foster City, California). All sequences were proofread manually.

Pharmacokinetics

Samples of venous blood (5 mL each) were collected into plastic tubes (heparin was used as an anticoagulant agent) from the first eight patients in each treatment group. Samples for pharmacokinetics analysis were taken as follows. On day 1 and day 112 (week 16), samples were taken before and 1, 2, 4, 5, 6, 8, 9, 10, and 12 hours after the first morning dose. On day 28, they were taken before and 1, 2, 4, 5, 6, 8, 9, 10, 12, 13, 14, 16, 17, 18, 20, 21, 22, and 24 hours after the first morning dose. Patients were given food before all doses were given.

The samples were kept on ice until the plasma was separated by centrifugation (which was done within 30 minutes of collection) and stored at − 20 °C. Before shipment, all samples were inactivated by heat treatment in a waterbath at 56 ± 1 °C. Samples were stored at − 20 °C before shipment. Total plasma concentrations were measured by radioimmunoassay using a method under the control of the pharmacokinetics and metabolism department at Roche Products Ltd. (Welwyn, United Kingdom). Individual pharmacokinetics variables were derived by “noncompartmental” methods. Maximum plasma concentration, time to maximum concentration, minimum plasma concentration, and time of minimum plasma concentration were observed variables. The area under the concentration versus the time curve was calculated (Topfit 2.0; Gustav, Fischer, and Verlag; New York, New York) by the trapezoidal method, assuming linear changes between points. The average concentration was calculated by dividing the area under the concentration curve by the period over which it was calculated. Accumulation was assessed by dividing the area under the concentration curve for the last dose on day 1 by the corresponding area under the concentration curve from the first dose on day 1.

Statistical Analysis

Statistical analysis was done by Bradley Efron. Analyses were done using one- or two-sample Wilcoxon tests, always two-sided, on log differences.

Results

Forty volunteers who were positive for HIV-1 and had CD4+ T-cell counts of 200 to 500 cells/mm³ were enrolled. Twenty received 600 mg of saquinavir (Roche Products, Welwyn, United Kingdom) six times daily for a total daily dose of 3600 mg, and 20 received 1200 mg of saquinavir six times daily for a total daily dose of 7200 mg. The study included 39 men and 1 woman who were 25 to 55 years of age (mean age, 39.7 years) and had an average CD4+ T-cell count of 346 cells/mm³ (range, 188 to 527 cells/mm³). The age distribution did not differ significantly between the treatment groups. The average CD4+ T-cell count in the low-dose group was 349 ± 86 cells/mm³ (range, 188 to 498 cells/mm³). This was not significantly different from the average CD4+ T-cell count in the high-dose group, which was 342 ± 95 cells/mm³ (range, 212 to 527 cells/mm³). Results of tests for p24 antigen were positive in 11 patients in the low-dose group and 3 patients in the high-dose group at study enrollment. The mean plasma RNA level at baseline was higher for the low-dose patients (4.86 ± 0.72 log RNA copies/mL of plasma [range, 3.21 to 5.87 log RNA copies/mL of plasma]) than for the high-dose patients (4.39 ± 0.79 log RNA copies/mL of plasma [range, 2.79 to 5.6 log RNA copies/mL of plasma]) (P = 0.053). Sixteen patients in each group had not had substantial previous antiretroviral therapy (< 3 months). Four patients in each group had had previous zidovudine therapy and had evidence of mutation at codon 215 of the reverse transcriptase gene, which is known to confer resistance to zidovudine. Three patients receiving the high-dose saquinavir regimen did not complete the study. One discontinued therapy during the first week; this patient reported lightheadedness, which resolved when he stopped taking saquinavir, but he refused any further follow-up. Thus, his data were not included. Patients receiving the high-dose regimen were required to take 36 pills each day, and 2 patients stopped receiving therapy prematurely because of poor compliance, one after 12 weeks and one after 16 weeks. Neither had evidence of toxicity. Their data up to the time that they dropped out of the study were included.

CD4+ T-cell Counts

The low-dose group had an initial increase in mean CD4+ T-cell count of 72 cells/mm³ (95% CI, 22.4 to 139.2 cells/mm³). The count peaked at week 4 and thereafter gradually decreased toward the baseline value: The mean CD4+ T-cell count at week 24 was 31 cells/mm³ (CI, minus9.1 to 70.5 cells/mm³) higher than the value at baseline (Figure 1). The mean elevation from baseline value was found to be statistically significant for weeks 2 through 16 (P < 0.04 for all comparisons). The mean elevation from baseline values at weeks 20 and 24 was not statistically significant (P > 0.19). Seventeen patients in this low-dose group had an initial increase in CD4+ T-cell count of at least 50 cells/mm³. The mean CD4+ T-cell count in the high-dose group increased to a peak of 121 cells/mm³ (CI, 68.1 to 166.9 cells/mm³) higher than the baseline value at week 20 and remained 82 cells/mm³ (CI, 26.3 to 113.5 cells/mm³) higher than the baseline value at week 24. The mean elevation from baseline values in this group was found to be statistically significant for weeks 2 through 24 (P < 0.02 for all comparisons). All patients in the high-dose group had an increase in CD4 count of at least 50 cells/mm³. The mean elevation in CD4 count from baseline values was significantly greater in the high-dose group than in the low-dose group at weeks 12 (P = 0.03) and 20 (P = 0.002). The difference between the groups did not reach statistical significance at the other time points.

Viral Load as Measured by Reverse Transcriptase Polymerase Chain Reaction for Plasma HIV RNA Levels

The mean plasma HIV RNA level decreased from 4.86 to 3.82 log RNA copies/mL of plasma in the low-dose group (a reduction of 1.06 log RNA copies/mL of plasma [CI, 0.76 to 1.25 log RNA copies/mL of plasma]) and from 4.39 to 3.05 log RNA copies/mL of plasma in the high-dose group (a reduction of 1.34 log RNA copies/mL of plasma [CI, 1.01 to 1.58 log RNA copies/mL of plasma]) (Figure 1). In the low-dose group, the mean plasma HIV RNA level showed a maximal decrease at week 2 and then gradually increased toward the pretreatment value, remaining 0.48 log RNA copies/mL of plasma (CI, 0.21 to 0.64 log RNA copies/mL of plasma) less than the baseline value at week 24. In the high-dose group, plasma HIV RNA levels declined to a nadir at week 4 and then gradually increased toward baseline values, remaining 0.85 log RNA copies/mL of plasma (CI, 0.37 to 1.10 log RNA copies/mL of plasma) less than the baseline value at week 24. The decrease in the mean plasma HIV RNA level to less than baseline was statistically significant for both groups throughout the entire study (P < 0.001), as was the difference in the magnitude of the decrease between the two groups (P < 0.02 for all comparisons except that at week 20; P = 0.08 for the comparison at week 20) at all time points. Fourteen patients receiving the low-dose regimen and 16 patients receiving the high-dose regimen had a maximal decrease of at least one log. At week 24, one patient in the low-dose group and 6 patients in the high-dose group had a sustained decrease of at least one log. This difference between the groups was statistically significant (P = 0.03 using the chi-square for a 2 × 2 table).

Additional Viral Load Markers

Eleven patients in the low-dose group and three patients in the high-dose group were positive for immune-complex-disassociated p24 antigen at baseline. Both groups showed a reduction in immune-complex-disassociated p24 antigen levels that lasted throughout the study (Figure 2). The maximal mean decrease, a decrease to 20% (CI, 4% to 88%) of the baseline value in the low-dose group and to 10% (insufficient sample for calculation of CIs) of the baseline value in the high-dose group, occurred at week 12. These maximal mean decreases from baseline reached borderline statistical significance (P = 0.06 for the low dose; P = 0.06 for the high dose). No statistically significant difference was seen in the reduction in immune-complex-disassociated p24 antigen levels between the high-dose and low-dose groups. Twenty patients in the low-dose group and 19 patients in the high-dose group had positive quantitative peripheral blood mononuclear cell dilution cultures at baseline and therefore could be monitored throughout the study. Culture titers decreased by a mean of 1.2 log DNA copies per million CD4 cells (CI, 0.42 to 1.73 log DNA copies per million CD4 cells) in the low-dose group and a mean of 1.6 log DNA copies per million CD4 cells (CI, 1.06 to 2.10 log DNA copies per million CD4 cells) in the high-dose group, with a nadir at week 8. Titers increased slightly thereafter but remained suppressed through week 24. The mean decrease in the high-dose group was significantly greater than that in the low-dose group at all time points (P < 0.03). Polymerase chain reaction for peripheral blood mononuclear cell viral DNA levels showed a slight reduction in both groups throughout the study. The mean maximal decrease occurred at week 16 and was 0.25 log DNA copies per million CD4+ T cells (CI, 0.10 to 0.39 log DNA copies per million CD4+ T cells) for the low-dose group and 0.41 log DNA copies per million CD4+ T cells (CI, 0.27 to 0.54 log DNA copies per million CD4+ T cells) for the high-dose group. The mean decrease from baseline was not statistically significantly different between the groups (P > 0.05 except at week 16; P = 0.03 at week 16).

The virologic response of the patients with the codon 215 mutation in their HIV-1 reverse transcriptase gene at baseline did not differ from that of the patients with the wild genotype.

Pharmacokinetics

The number of hemolyzed or clotted samples made it impossible to calculate pharmacokinetics data for one patient on day 1 and one patient on day 112 (both patients were in the high-dose group). In addition, data from one patient at day 112 had not been analyzed at the time of pharmacokinetics assessment. Summary statistics were calculated excluding the data obtained from these patients on the specified days. The mean values of the pharmacokinetics variables are shown in Table 1. Because of the short dose interval and the long absorption period of saquinavir, it was not possible to discern classic pharmacokinetics profiles for separate dose intervals, and pharmacokinetics variables were calculated over 12 or 24 hours as appropriate. The number of patients in whom pharmacokinetics data were obtained was too small to allow for statistical comparisons between groups.

In both groups, plasma concentrations increased rapidly on day 1, and maximum concentrations were reached at 12 hours. By 12 hours, concentrations had increased 4.5-fold in the low-dose group and 3-fold in the high-dose group. The mean plasma concentration on both day 28 (Figure 3) and day 112 increased toward the end of the sampling period. For both the low-dose and high-dose groups, the mean values for the area under the curve to 24 hours on day 28 were twice the areas under the curve calculated over the first 12 hours on day 112 (Table 1). Plasma concentrations over the first 12 hours on days 28 and 112 were similar in the two groups, as were the mean concentrations on days 28 and 112 (Table 1). The maximum concentrations were higher on day 28 because higher concentrations were seen toward the end of the 24-hour period in this group. Mean plasma concentrations for the high-dose group were more than twice those seen for the low-dose group on all sampling days. The mean value for the area under the curve to 24 hours on day 28 was approximately four times higher for the high-dose group (9742 ng · h per mL [48%]) than for the low-dose group (2233 ng · h per mL [35%]). The maximum concentrations achieved on the sampling days were between four and seven times higher for the high-dose group than for the low-dose group.

Drug concentrations were correlated with decrease in viral load for patients who had pharmacokinetics studies (Figure 4). For each patient, the area under the curve at week 4 was plotted against the decrease in plasma HIV RNA level from baseline at week 4. A strong direct correlation was found between drug concentrations and the decrease in viral load from baseline. This was found for both treatment groups and for the data from both groups combined.

Mutations

Mutations at codons 48 and 90 of the HIV protease gene were determined in plasma samples from each patient by using the selective PCR technique at baseline and at week 24 (Table 2). Samples that showed mutations at week 24 were re-tested at earlier time points to determine the time at which the mutation had first appeared. All patients in both treatment groups had wild-type sequences at both codon 48 and codon 90 at baseline. In the low-dose group, two patients had developed mutations at codon 48 and seven had developed mutations at codon 90 by week 24. In the high-dose group, two patients had developed mutations at codon 48 and two had developed mutations at codon 90. No patients developed both mutations. Altogether, nine patients in the low-dose group and four patients in the high-dose group had developed one of the mutations by the end of the study. This difference did not reach statistical significance (P = 0.09 for the chi-square test). The more sustained suppression of viral load seen in the high-dose group appears to have been accompanied by a lower mutational rate, whereas the weaker suppression of viral load seen in the low-dose group appears to have been accompanied by a higher mutational rate (Figure 5).

Complete Applied Biosystems sequencing of the entire protease gene was done on a subset of 38 samples. A comparison of the results of the Applied Biosystems sequencing and of the PCR assay in the 38 samples tested by both assays showed that there was a 94% correlation between the techniques in the identification of wild-type or mutant genotype at codons 48 and 90. The only discrepancies appeared with the PCR assay showing greater sensitivity in the early detection of mutations as had been seen in our other investigations. Of the 19 samples at week 24 on which complete sequencing of the entire protease gene was done, 1 had acquired the codon 54 mutation (I→V). This patient had also acquired the codon 48 mutation. None of these samples showed the mutations at codons 82 (V→T) or 84 (I→V).

Toxicity

Three patients in the high-dose group and one in the low-dose group required drug “holidays” of 4 to 14 days because of adverse reactions. One had nausea and diarrhea, one was a hepatitis B surface antigen-positive patient and had elevated serum aminotransferase levels, one had increased neutropenia after having a baseline absolute neutrophil count of 1050 cells/mm³, and one had elevated serum creatinine phosphokinase levels (as high as 18 514 U/L) after initiation of a strenuous physical exercise program (Table 3). All toxicities resolved when therapy was discontinued. Therapy was restarted at the full high dose, and the creatinine phosphokinase toxicity did not recur. The elevated serum aminotransferase level recurred after 4 months and required an additional 1-week drug holiday, after which the patient received the full dose for an additional 4 months and had aminotransferase levels in the high end of the normal range. The absolute neutrophil count of the patient with neutropenia returned to the baseline value after the drug holiday, and the patient required a second holiday after 10 weeks. Seven patients in the low-dose group and four in the high-dose group reported mild confusion or lightheadedness that resolved within 1 to 2 weeks in all but one patient.

Physical examination showed no objective findings in these patients, and a side effect interfered with normal function in only 1 patient. Eleven patients in the low-dose group and 14 patients in the high-dose group had gastrointestinal toxicity, which predominantly presented as diarrhea with occasional abdominal pain and nausea. In 2 patients in the low-dose group and 4 patients in the high-dose group, the toxicities were characterized as moderate; in the other patients, the toxicities were considered mild. Most gastrointestinal toxicities occurred in the first 1 to 2 weeks of therapy and decreased or disappeared thereafter without dose modification.

Drug Continuation Phase after the Study

Patients who showed 1) a sustained increase in CD4+ T-cell counts of 50 cells/mm³ or more or 2) a sustained suppression of plasma HIV RNA level of at least 0.5 log RNA copies/mL of plasma less than baseline, or both, throughout the entire study were allowed to continue receiving saquinavir at the designated dosage after the study had ended. After completing the study, 10 of 20 patients in the low-dose group and 15 of 17 patients in the high-dose group fulfilled these criteria and chose to continue receiving study medication through week 48. At week 32, 9 patients in the low-dose group (45% of the original group) still fulfilled the criteria for sustained response. Two patients in the high-dose group have not yet reached week 32. Of the 15 who have, 12 (75%) still fulfill the criteria for sustained response.

Discussion

We examined the efficacy, safety, and pharmacokinetics of saquinavir therapy at twice and fourfold the dose currently licensed for use (1800 mg/d). Both of the doses we studied increased CD4+ T-cell counts and reduced plasma HIV RNA levels for the 24 weeks of our study. Additional virologic markers, including peripheral blood mononuclear cell cultures, immune-complex-disassociated p24 antigen levels, and peripheral blood mononuclear cell viral DNA levels were all reduced as a result of therapy. The increase in CD4+ T-cell counts and the reduction in plasma HIV RNA levels was greater and lasted longer with the high dose (7200 mg/d) than with the low dose (3600 mg/d). The additional virologic markers also showed a greater reduction with the higher dose. Because both doses continued to decrease these markers at 24 weeks, no difference between the doses in the duration of effect could be determined. A longer observation period may be needed to show that the high-dose regimen has a more durable effect on these virologic markers. When comparing our results with those of a published clinical trial that studied saquinavir monotherapy at doses as high as 1800 mg/d [7], caution must be used, because the two studies differed in baseline patient characteristics, virologic assay techniques, and data analysis, all of which can greatly affect study results. However, even with these limitations, there still appears to be a larger and longer improvement in both plasma HIV RNA levels and CD4+ T-cell counts with higher doses of saquinavir. The dose of 1800 mg/d resulted in a maximal median reduction in plasma HIV RNA levels of about 0.7 log RNA copies/mL of plasma (a reduction of 80%). The low dose (3600 mg/d) in our study resulted in a maximal mean reduction of 1.06 log RNA copies/mL of plasma, and the high dose (7200 mg/d) resulted in a maximal mean reduction of 1.34 log RNA copies/mL of plasma. The duration of viral suppression also seemed to increase with higher doses: Plasma HIV RNA levels had returned to baseline by week 16 with the 1800-mg dose but continued to be 0.48 and 0.85 log RNA copies/mL of plasma less than baseline values at week 24 with the 3600-mg and 7200-mg doses, respectively. Similarly, the maximal increase in the median CD4+ T-cell count was 50 cells/mm³ in the study of the 1800-mg dose, and, in our study, the mean maximal increase was 72 cells/mm³ with the 3600-mg dose and 121 cells/mm³ with the 7200-mg dose. Here, too, the duration of the effect on CD4+ T-cell counts was greater with the higher doses. Of note, in the study of the 1800-mg dose, doses of 75 to 600 mg/d were also studied, and the results showed an increasing effect with increasing doses of saquinavir between 75 and 1800 mg/d. Our findings are also consistent with those of two recently published studies that showed that another HIV protease inhibitor, ritonavir, had a greater and more durable effect with increasing doses [32, 33]. The toxicity profile seen with high doses of saquinavir was favorable compared with those of the dideoxy nucleotide reverse transcriptase inhibitors licensed to date [34-36]. Although the severity and frequency of adverse effects were greater in the high-dose group than in the low-dose group, most patients in the high-dose group had minimal or no toxicity and could complete the study without dose modification. The few major adverse reactions seemed to develop when high doses were administered in the presence of an additional contributing host factor. For example, neutropenia developed in a patient with a suppressed absolute neutrophil count at baseline, and an elevated serum creatinine phosphokinase level appeared in a patient after the patient had started a new, strenuous physical exercise program. It is important to note that all such adverse events were reversible and rapidly cleared with brief discontinuation of therapy.

Clinical studies that used lower doses of saquinavir also reported similar clinical side effects and elevations in serum aminotransferase and creatinine phosphokinase levels. Most toxicities were graded as mild or moderate, and almost none were severe [7]. When used as a component of combination therapy at a dose of 1800 mg/d, saquinavir also had a favorable toxicity profile; the combination of zidovudine, zalcitabine, and saquinavir had no more toxicity than the combination of zidovudine and zalcitabine alone [23].

Our pharmacokinetics results show that saquinavir has modest accumulation, with median increases of 4.5-fold for the low-dose regimen and 3.0-fold for the high-dose regimen. These are slightly greater than the increases seen in earlier studies, but this may be because of the small amount of data available in our study. Once steady state was reached, no change was evident in the pharmacokinetics of saquinavir over the 16-week period; the concentration-time profiles and the average concentration values were similar for both day 28 and day 112. The increase in plasma concentrations of saquinavir over the sampling interval on days 24 and 112 may have been caused by the hospitalization of the patients over the sampling period, which improved compliance. This, however, makes the detection of any circadian effect impossible. The increase in exposure (as represented by the values for the area under the concentration curve to 24 hours) between the two treatment groups is greater than would have been expected from the increase in dose. This is probably caused by the saturation of the enzymes responsible for the metabolism of saquinavir during the absorption process. Saquinavir is rapidly metabolized and undergoes extensive first-pass metabolism, some of which may take place in the gut wall as well as in the liver. Saturation of metabolism in the gut wall or the liver could occur at the very high concentrations seen during absorption. The greater effect of the higher dose on surrogate markers is supported by the finding that higher area-under-the-curve values correlated with a greater decrease in plasma HIV RNA level (Figure 4).

Nine patients receiving the low-dose regimen and four patients receiving the high-dose regimen developed key resistance mutations to saquinavir by week 24. Mutations also tended to appear later in the high-dose group than in the low-dose group. Although the difference between the mutation rates did not reach statistical significance, possibly because of the small numbers, the larger and longer suppression of viral load induced by the higher dose may delay the appearance of resistance mutations. This has also been suggested by others [32] as a possible mechanism for the increased efficacy of higher doses of ritonavir. This suggests that HIV protease inhibitors are unique and contrasts with observations reported in patients who were receiving dideoxy nucleoside reverse transcriptase inhibitors that had less effect on viral load, such as zidovudine and didanosine. In these patients, higher doses actually appeared to result in higher rates of mutation detection [37-39]. Certain combinations of nucleosides capable of producing a suppression of viral load similar to that seen in our high-dose group have delayed the detection rate of specific resistance mutations [40, 41]. However, the exact mechanism of the relation between higher protease inhibitor doses, increased plasma viral suppression, and the apparent low mutation detection rate is unclear. The reduction in viral load, viral turnover, or both may decrease the opportunity for mutations to develop, or the high dose may affect the growth of the resistant mutant forms. We cannot rule out that other factors in our study, such as the lower baseline plasma HIV RNA level in the high-dose group, may have contributed to the apparent delay in the development of mutations.

Cross-resistance between protease inhibitors has been a concern, potentially limiting the use of other agents after patients have been treated with one drug of this class [9, 20-22]. Although some studies have found that resistance mutations to saquinavir largely do not overlap with resistance mutations to other protease inhibitors [22], other reports have found that cross-resistance to saquinavir emerges after patients receive therapy with other drugs [20]. The clinical relevance of these observations has yet to be determined. In our study, the small subset of samples studied by complete genotypic sequencing did not show the mutations at codons 82 (V→T) or 84 (I→V), which have been reported to emerge in vivo and produce cross-resistance to multiple protease inhibitors [20].

Our study has several limitations. Investigators and patients were not blinded to dose assignment, and no placebo group was included. Like many other early-phase studies on new antiretroviral agents, our study provides data not on truly clinical end points, such as mortality, but on surrogate markers, such as CD4+ T-cell count and plasma HIV RNA levels. Although these markers have been shown to correlate with disease progression and are commonly used to determine drug efficacy, these limitations must be kept in mind when our results are extrapolated to clinical practice.

Our results suggest that doses of saquinavir greater than that currently licensed may produce a larger and longer antiviral effect. Even at these higher doses, the toxicity profile of saquinavir appears to be acceptable. The development of resistance mutations may be delayed as a result of higher doses and a more profound and durable suppression of viral load. Preliminary data have shown that the addition of saquinavir at the licensed dose to combination therapy increases the efficacy of reverse transcriptase inhibitors without increasing toxicity. We believe that our results warrant further study of higher doses of saquinavir as monotherapy and, in particular, in combination with other antiretroviral agents. A new formulation of saquinavir with improved bioavailability that is currently being tested could result in higher exposures to the compound, which could lead to a reduction in the cost and number of pills taken daily.

Mr. Winters: Division of Infectious Diseases, Room S-146, 300 Pasteur Drive, Stanford University School of Medicine, Stanford, CA 94305.

Dr. Stewart: Roche Products Ltd., PO Box 8, Broadwater Road, Welwyn Garden City, Herts AL7 3AY, United Kingdom.

Dr. Efron: Health, Research, and Policy, Sequoia Hall, Room 106, Stanford University, Stanford, CA 94305-4065.

Ms. Norris: Center for AIDS Research, Room S-168, 300 Pasteur Drive, Stanford University School of Medicine, Stanford, CA 94305-5107.

Dr. Kozak Division of Infectious Diseases, 200 Hawkins Drive, The University of Iowa College of Medicine, Iowa City, IA 52242-1081.

Dr. Merigan: Center For AIDS Research, S-156 Grant Building, Stanford University, Stanford, CA 94305.