Table H1 provides a synopsis of the design features of the 4 trials. The first trial (R89) involved AIDS patients with CMV. The second trial involved HIV+ patients some of whom also had HBV. The 3rd trial involved previously untreated (in the sense of exposure to FIAU or FIAC) patients with chronic HVB. The fourth trial involved HVB patients, the majority of whom had been previously treated with FIAU (11 of the 15 enrolled). None of the trials involved a randomized control group, hence, statistical assessments of the results observed must be based on the assumed background mortality rates applying to the populations under study. Three hypothetical death rates (proportion of the patients expected to die each year) were used for each of the analyses that follow:

Low 
Int 
High 
R89 
0.10 
0.15 
0.20 
R90 
0.04 
0.08 
0.12 
R91 
0.02 
0.03 
0.04 
PPPC 
0.02 
0.03 
0.04 
These rates correspond to the median life expectancies in years listed below. The median life expectancy, assuming a constant mortality rate, is 0.7 divided by the rate (Kalbfleisch and Prentice, 1980). The corresponding standard error for the rate is the rate divided by the square root of the number of events observed.

Low 
Int 
High 
R89 
7.0 
4.7 
3.5 
R90 
17.5 
8.75 
5.83 
R91 
35.0 
23.3 
17.5 
PPPC 
35.0 
23.3 
17.5 
Below are the first 10 and last 10 pages of uncorrected machineread text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapterrepresentative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 256
Review of the Fialuridine (FIAU) Clinical Trials
H Statistical Analysis of Mortality in the FIAU/FIAC Clinical Trials
Table H1 provides a synopsis of the design features of the 4 trials. The first trial (R89) involved AIDS patients with CMV. The second trial involved HIV+ patients some of whom also had HBV. The 3rd trial involved previously untreated (in the sense of exposure to FIAU or FIAC) patients with chronic HVB. The fourth trial involved HVB patients, the majority of whom had been previously treated with FIAU (11 of the 15 enrolled). None of the trials involved a randomized control group, hence, statistical assessments of the results observed must be based on the assumed background mortality rates applying to the populations under study. Three hypothetical death rates (proportion of the patients expected to die each year) were used for each of the analyses that follow:
Low
Int
High
R89
0.10
0.15
0.20
R90
0.04
0.08
0.12
R91
0.02
0.03
0.04
PPPC
0.02
0.03
0.04
These rates correspond to the median life expectancies in years listed below. The median life expectancy, assuming a constant mortality rate, is 0.7 divided by the rate (Kalbfleisch and Prentice, 1980). The corresponding standard error for the rate is the rate divided by the square root of the number of events observed.
Low
Int
High
R89
7.0
4.7
3.5
R90
17.5
8.75
5.83
R91
35.0
23.3
17.5
PPPC
35.0
23.3
17.5
OCR for page 256
Review of the Fialuridine (FIAU) Clinical Trials
The labels low, intermediate, and high are labels of convenience for the presentation that follows. They are not to be construed as qualitative labels for the presumed actual underlying mortality rates for the populations of interest. The true underlying rates may, in reality, be higher than the levels indicated for the set carrying that label. For example, the median life expectancy for the population enrolled into R89 (AIDS patients with CMV) is, in all probability, less than the 3.5 years listed for the ''high'' rate. The same is likely to be true for the population studied in R90. The estimated median life expectancy is around 10 years from conversion. The population studied is likely to have been positive for some time and, hence, has a life expectancy considerably less than that.
The data used for the reanalyses are as given in Table H2, as derived from the FDA report. The closing date for recruitment into the individual trials is assumed to correspond to the date of the last dose for the trials, as recorded in the report. The period of recruitment is assumed to be that interval from enrollment of the first patient into a trial and the date of the last dose. The rate of enrollment was assumed to be constant over that interval. The actual period is less than that used herein by a matter of days or a few weeks, depending on the trial.
For the analyses presented below to be informative on a real time basis, one would have to have close surveillance of each study person so as to learn of deaths as they occur or within a matter of days after they occur. Lags in detection or reporting of deaths to the analysis site would degrade the value of the analyses as a real time decisionmaking tool.
Similarly, to be useful, one has to have 100% surveillance of all persons enrolled for mortality, regardless of their status in the trial (i.e., such surveillance including dropouts). We have no way of knowing the extent to which deaths noted in the FDA are the result of such surveillance. The report does not indicate the methods used to identify deaths occurring after the defined period of treatment and followup for the trials discussed. Hence, we have no way of knowing the extent to which the surveillance can be viewed as complete. There is, in fact, intrinsic evidence to suggest it is incomplete simply from the counts of deaths provided. For example, one would expect most of the patients in R89 to have died by mid 1993. The median life expectancy for the population enrolled into that trial was not likely to be much more than a year, if that.
The caveats regarding followup need to be kept in mind when reviewing the results below. An underreporting of deaths would mean that the actual pvalues noted in the tables below would be smaller than given and that the powers would be larger than indicated. Since the ability to trace people diminishes as a function of the time of the last contact, we surmise that the quality of followup was better for the 26 week period of followup than for followup through 27 June 1993. Hence, the emphasis is on analyses relating to 26 weeks of followup.
The results for each trial analyzed separately is given in Tables H3, 4, and 5 for the low, intermediate, and high rates. The first panel of each table is at the close of each trial assuming 26 weeks of followup for each person enrolled. Note that for this method of analysis only 4 of the 9 deaths observed in the first three trials are counted. The other 5 deaths occurred beyond the closing date for followup and are censored. The second panel is for followup through 27 June 1993.
The sample size, assuming a uniform rate of enrollment over the interval defined by enrollment of the first patient to the last administration of treatment in a given trial is given in the column labeled N.
OCR for page 256
Review of the Fialuridine (FIAU) Clinical Trials
Expected deaths are calculated assuming the specified event rate for a table or a panel within a table. The total number of expected person years contributed by a total of N persons followed to death or to the indicated end of followup, is given in the fifth column of the tables 3, 4, 5 and 6. For example, it is 5.40 person years for R89 assuming 26 weeks of followup (i.e., 0.90 times 6.0). The expected number dead is 0.60 (i.e., 6.0 times 0.10). The expected deaths in table 6 correspond to the sum of expected deaths at the time of a death. For example, the value of 1.700 at the start of PPPC for first panel of table 6 is the sum of 0.60 + 0.86 + 0.24 (for R89, R90, and R91, respectively).
The pvalues are for differences between the observed and expected numbers of deaths, and are derived using a technique described by Breslow and Day (1987) involving standardized mortality rates and the Poisson distribution. The pvalues are for a twotailed alternative. The calculations are made assuming the comparison of two groups, each having the sample size of that for the indicated trial.
The direction of the observed treatment effect is determined by the sign of the difference resulting from subtracting expected deaths from observed deaths (column 9 in Table Hs 3, 4, and 5, column 6 in Table H6). A positive difference corresponds to an excess (measured against the number expected given the underlying mortality rate) of deaths associated with the test treatment. A negative difference corresponds to fewer than the number of deaths expected. The latter difference is indicative of a positive treatment effect and the former is indicative of a negative treatment effect. For the 3 trials done prior to PPPC and for the 26 weeks of followup, two were indicative of varying degrees of a negative treatment effect (R89 and R91) and one of varying degrees of a positive effect (R90). A positive treatment effect is one where the expected number of deaths is larger than the number actually observed and a negative treatment effect is one where the reverse is true.
Powers are presented in the last two columns of Tables H3, 4, and 5 for a specified negative and positive alternative. Power is the probability of rejecting the null hypothesis, that the test treatment has no effect on mortality, when, in fact, it has the indicated negative or positive effect. Powers of 0.80 or higher are desired when designing randomized controlled trials with a calculated sample size. The power calculations are made with the type I error (a) fixed at 0.05 and correspond to powers calculated prior to the start of the trials.
The negative alternative considered is that the test treatment produces a 100% increase in mortality and the positive alternative is that it produces a 50% reduction in mortality. For example, if the assumed background mortality rate is 0.10, than the negative alternative is that the actual rate, in the presence of the test treatment, is 0.20 instead of the of the 0.10 hypothesized and that it is 0.05 instead of the 0.10 hypothesized if the effect of the test treatment is positive. In the case of R89, for 6 months of followup for a presumed low underlying mortality rate of 0.10 (Table H3), the calculated power prior to the start of the trial against the two alternatives, given a sample size of 12 for both the test and control treated groups, is 0.08 and 0.07, respectively. A power of 0.08 means that, given the sample size of 12 per treatment group, the trial would have only about 1 chance in 12 (8 in 100) of detecting a 50% true positive effect of the drug and about the same chance of detecting a 100% negative effect of the drug, given a hypothesized underlying rate of 0.10 for the control treated group.
Not surprisingly, given the size of the trials, all of the powers are low. Given the relatively nominal nature of the observed difference and the resulting pvalues, the calculated
OCR for page 256
Review of the Fialuridine (FIAU) Clinical Trials
powers suggest the need for a good deal of skepticism in concluding for or against the test treatment based on any single trial. In other words, one runs the risk of drawing the wrong conclusion regarding the treatment, even when the observed difference in a trial is indicative of harm.
The only trial that produces, according to a criterion based on pvalues, any convincing evidence of harm is PPPC, and all of that information accumulated after the investigators had already taken action, based on morbidity, to stop the trial.
Even if one were to assume the form of analysis described in Table H6, combining data from all trials in cumulative fashion, the information against the drug prior to the deaths in PPPC is weak. The smallest pvalues are after the third death. The pvalues observed at the start of PPPC are unimpressive, ranging in size from 0.19 to 0.77 for differences (observed minus expected deaths) ranging from 2.300 to 0.260.
The approach for Table H6 is that of an analyst performing a new analysis each time another death is observed over the course of the 3 trials. The three panels in the Table are for 26 weeks of followup, with censoring occurring at the start at the 27th week after enrollment of a patient. The personyears recorded in Table H6 are the sums of those available from the three trials at the indicated point of analysis. For example, the maximum amount available for 26 weeks of followup at the death of patient 4D, 6 Jan 1993, was 6.000 person years for R89, 21.500 for R90, and 11.099 for R91, for a total of 38.6 person years. (The study still generating information at 26 weeks of followup at that point in time was R91. R89 and R90 would have completed 26 weeks of followup on patients enrolled in those trials prior to 6 Jan 1993. There were 3 deaths observed within the defined period of followup prior to the one in question, for a total of 4 at the point of analysis.) The expected number of deaths is the sum of the products of person years of experience from a specific trial multiplied by the presumed underlying mortality rate for that study population; for example, 0.10(6.0) + 0.04(21.5) + 0.02(11.1) for a total of 1.68.
The analyses summarized in Table H6 are for deaths occurring within 26 weeks of enrollment. Deaths beyond that point are not counted. The analyses do not take account of the redosing of 4 patients in R90 (401, 406, 408, and 409) or of the 3 deaths occurring among them them (401,406, and 408) because the redosing and deaths occurred after their respective 26th weeks of followup. A different approach is needed to deal with all deaths regardless of time from enrollment. That approach is represented in Table H7.
The metaanalytic approach underlying Table H7 is akin to that which could be used if investigators, just prior to starting their next trials, were to perform an analysis using all data accumulated up to that point to help them decide whether they should proceed to the next trial. The smallest pvalue indicated of possible harm is the one for low underlying mortality rates for data accumulated just prior to the start of R90. The smallest pvalue indicative of possible benefit is the one for high underlying mortality rates for data accumulated just prior to the start of PPPC.
The analyses are instructive only to the degree that one has faith in the underlying morality rates used and in the validity of the assumptions made, and among those, principally the one having to do with completeness of followup. The likelihood is that we have been conservative in the range of morality rates investigated. Hence, "low" may be too low and "high" may not be high enough. Obviously, any increase in assumed underlying rates weakens
OCR for page 256
Review of the Fialuridine (FIAU) Clinical Trials
the case for harm known in advance of PPPC. Conversely, the analysis is, in effect, a "best case" analysis in regard to completeness of followup for deaths. The analyses performed assume 100 percent ascertainment for deaths. Under that assumption, persons not located when the FDA task force report was complied are assumed to have been alive. The report does not contain details on completeness of followup. Hence, we could not perform analyses under other assumptions regarding patients not located at the time the report was complied.
Neither of the analyses in Tables H6 or 7 take account of actual amount of drug administered and, hence, are not helpful in addressing the broader question as to whether the studies taken together are indicative of increasing probability of death with increased exposure to the drug. The relative risk (per unit increase of dose on the log scale) are 2.6 and 4.6, respectively, for censoring of followup at 26 weeks after enrollment (4 deaths) and March 24, 1993, (9 deaths; the point just prior to the start of PPPC), as obtained via Cox proportional hazards analyses (Cox, 1972). The pvalues corresponding to the relative risks resulting from the two forms of censoring were 0.18 and 0.013, respectively. To be sure, all analyses involving dosage administration are problematic to the extent that dosage is a function of other intervening variables. Hence, the interpretation given to these results, even differences yielding a pvalue as small as the one for the latter form of censoring, has to be treated with caution.
Tables H1 through H7 follow.
OCR for page 256
Review of the Fialuridine (FIAU) Clinical Trials
Table H1. Design Features of Individual Trials
R89001
R90001
R91010
H3XPPPC
Phase
I/II
Dose escalation
I/II
Dose escalation
I/II
Dose escalation
I/II
Treatment
Randomized control
None
None
None
None
IND
Oclassen 33,638
Oclassen 34,973
Oclassen 34,973
Lilly
No. of centers
2
3
1
1
Investigators
Richman Corey
Richman Corey Straus
Hoofnagle
Hoofnagle
Performance sites
U. of Cal. S.D.
U. of Wash., Seattle
U. of Cal., S.D.
U. of Wash., Seattle
NIH, Bethesda
NIH, Bethesda
NIH, Bethesda
Drug
FIAC
FIAU
FIAU
FIAU
Treatment period
28 days
14 days
28 days
168 days
1st dose
17 Nov 89
9 Oct 90
13 Apr 92
24 Mar 93
Last dose
25 May 90
29 Jun 92
10 Sep 92
28 Jun 93
No. enrolled
12
43
24
15
Enrollment period
188 days
628 days
150 days
95 days
Enrollment rate
15.67 days
6.05 days
6.25 days
6.33 days
No. of Deaths
4
4
1
5
Prior FIAC/FIAU
No
No
No
Yes (11)
Study pop.
AIDS (2)
AIDS+CMV (10)
HIV (13)
HIV+HVB (30)
Compensated HVB
Compensated HVB
Treatment and followup
35 days
42 days
168 days
168 days
Dose
Low
0.6 mg/kg/day
0.1 mg/kg/day
0.05 mg/kg/day
0.10 mg/kg/day
High
1.0 mg/kg/day
1.7 mg/kg/day
0.5 mg/kg/day
0.25 mg/kg/day
Project titles:
R89001: Efficacy and safety of oral FIAC in AIDS patients with cytomegalovirus infection: A doseranging study.
R90001: The tolerance of HIVinfected patients with herpesgroup infections to oral doses of FIAU.
R91010: FIAU oral doseranging study in patients with compensated chronic hepatitis B.
H3XMCPPPC: FIAU redosing for the treatment of patients with compensated chronic hepatitis B: a sixmonth course.
OCR for page 256
Review of the Fialuridine (FIAU) Clinical Trials
Table H2. Death Experience by Calendar Time
Event
PI
Cum. N
Deaths
Cum. % Dead
Study Time
17 Nov 89
1st dose R89001
Richman,
1
10 Jan 90
103 enrolled
6 Feb 90
105 enrolled
22 Feb 90
107 enrolled
15 Mar 90
110 enrolled
15 Apr 90
105 died
1
10 wks
18 May 90
110 died
2
9 wks
25 May 90
Last dose R89001 (n=12)
12
16.7
2 June 90
107 died
3
17 wks
9 Oct 90
1st dose R90001
Richman,
13
11 Mar 91
103 died
Corey,
4
60 wks
16 Apr 91
401 enrolled
Straus
29 May 91
101 enrolled
4 Jun 91
406 enrolled
3 Nov 91
408 enrolled
13 Apr 92
1st dose R91010
Hoofnagle
56
16 May 92
408 died
5
27 wks
29 Jun 92
Last dose R90001 (n=43)
55
9.1
5 Aug 92
4D enrolled
11 Aug 92
406 died
6
62 wks
2 Sep 92
101 died
7
65 wks
10 Sep 92
Last dose R91010 (n=24)
79
10.1
18 Oct 92
401 died
8
78 wks
6 Jan 93
4D died
9
22 wks
24 Mar 93
1st dose H3XMCPPPC
Hoofnagle
80
24 Mar 93
0012002 enrolled
24 Mar 93
0013001 enrolled
7 Apr 93
0016003 enrolled
14 Apr 93
0013004 enrolled
15 Apr 93
0015003 enrolled
28 Jun 93
H3XMCPPPC stopped (n=15)
94
10.4
5 Jul 93
0012002 died
10
14 wks
6 Jul 93
0013004 died
11
13 wks
16 Jul 93
0015003 died
12
14 wks
30 Jul 93
0013001 died
13
19 wks
31 Aug 93
0016003 died
14
17 wks
14.9
OCR for page 256
Review of the Fialuridine (FIAU) Clinical Trials
Table H3. Results per Individual Trial for Low Underlying Mortality Rates
Personyrs
Deaths
Power
Rate
N
Avg FU
Exp
Obs
Exp
Obs
Diff
pvalue
Neg
Pos
26 week followup
R 89
0.10
12
0.50
5.40
5.19
0.60
3
2.40
0.05
0.08
0.07
R 90
0.04
43
0.50
20.64
21.50
0.86
0
0.86
—
0.13
0.09
R 91
0.02
24
0.50
11.76
11.92
0.24
1
0.76
0.43
0.06
0.06
Total
79
0.50
37.80
38.61
1.70
4
2.30
0.19
27 June 93 followup
R 89
0.10
12
3.58
38.66
30.49
4.30
4
0.30
0.75
0.19
0.13
R 90
0.04
43
1.50
61.92
62.97
2.58
4
1.42
0.52
0.20
0.12
R 91
0.02
24
1.33
31.28
31.01
0.64
1
0.36
0.94
0.08
0.06
PPPC
0.02
15
0.13
1.91
1.95
0.04
0
0.04
—
0.08
0.06
Total
94
1.50
133.77
126.42
7.56
9
1.44
0.69
Assumptions and notes: Enrollment period from 1st dose to last dose; uniform rate of enrollment over that period; instantaneous treatment effect (ie, no treatment lag); instantaneous reports of deaths; 100% followup for mortality over defined followup period; assumed underlying annual mortality rates: R89, 0.10; R90, 0.04; and R91, 0.02.
Table H4. Results per Individual Trial for Intermediate Underlying Mortality Rates
Person
Deaths
Power
Rate
N
Avg FU
Exp
Obs
Exp
Obs
Diff
pvalue
Neg
Pos
26 week followup
R 89
0.15
12
0.50
5.10
5.19
0.90
3
2.10
0.12
0.10
0.07
R 90
0.08
43
0.50
19.78
21.50
1.72
0
1.72
—
0.22
0.13
R 91
0.03
24
0.50
11.64
11.92
0.36
1
0.64
0.60
0.07
0.06
Total
79
0.50
36.52
38.61
2.98
4
1.02
0.70
27 June 93 followup
R 89
0.15
12
3.58
36.52
30.49
6.44
4
2.44
0.23
0.24
0.16
R 90
0.08
43
1.50
59.34
62.97
5.16
4
1.16
0.49
0.32
0.20
R 91
0.03
24
1.33
30.96
31.01
0.96
1
0.04
1.00
0.09
0.07
PPPC
0.03
15
0.13
1.89
1.95
0.06
0
0.06
0.24
0.05
0.05
Total
94
1.50
128.71
126.42
12.62
9
3.62
0.24
Assumptions and notes: Enrollment period from 1st dose to last dose; uniform rate of enrollment over that period; instantaneous treatment effect (i.e., no treatment lag); instantaneous reports of deaths; 100% followup for mortality over defined followup period; assumed underlying annual mortality rates: R89, 0.15; R90, 0.08; and R91, 0.03.
OCR for page 256
Review of the Fialuridine (FIAU) Clinical Trials
Table H5. Results per Individual Trial for High Underlying Mortality Rates
Personyrs
Deaths
Power
Rate
N
Avg FU
Exp
Obs
Exp
Obs
Diff
pvalue
Neg
Pos
26 week followup
R 89
0.20
12
0.50
4.80
5.19
1.20
3
1.80
0.24
0.11
0.08
R 90
0.12
43
0.50
18.92
21.50
2.58
0
2.58
—
0.29
0.18
R 91
0.04
24
0.50
11.52
11.92
0.48
1
0.52
0.76
0.07
0.06
Total
79
0.50
35.24
38.61
4.26
4
0.26
0.77
27 June 93 followup
R 89
0.20
12
3.58
34.37
30.49
8.59
4
4.59
0.06
0.27
0.19
R 90
0.12
43
1.50
56.76
62.97
7.74
4
3.74
0.10
0.43
0.26
R 91
0.04
24
1.33
30.64
31.01
1.28
1
0.28
0.56
0.10
0.08
PPPC
0.04
15
0.13
1.87
1.95
0.08
0
0.08
—
0.10
0.08
Total
94
1.50
123.64
126.42
17.69
9
8.69
0.02
Assumptions and notes: Enrollment period from 1st dose to last dose; uniform rate of enrollment over that period; instantaneous treatment effect (ie, no treatment lag); instantaneous reports of deaths; 100% followup for mortality over defined followup period; assumed underlying annual mortality rates: R89, 0.20; R90, 0.12; and R91, 0.04.
Table H6. Analysis on the Occurence of Each Death Based on 26 Weeks of Followup from Enrollment
Enrolled
Deaths
N
Nyrs
Obs
Exp
Diff
pvalue
26 week followup
Low Mortality (R89 0.10; R90 0.04; R91 0.02)
105
10.6
1.931
1
0.193
0.807
0.35
110
11.5
2.881
2
0.288
1.712
0.07
107
12.0
3.230
3
0.323
2.677
0.01
4D
79.0
36.920
4
1.648
2.352
0.17
PPPC
79.0
39.500
4
1.700
2.300
0.19
Intermediate Mortality (R89 0.15; R90 0.08; R91 0.03)
105
10.6
1.931
1
0.290
0.710
0.54
110
11.5
2.881
2
0.432
1.568
0.14
107
12.0
3.230
3
0.485
2.515
0.03
4D
79.0
36.920
4
2.903
1.097
0.66
PPPC start
79.0
39.500
4
2.980
1.020
0.70
High Mortality (R89 0.20; R90 0.12; R91 0.04)
105
10.6
1.931
1
0.386
0.614
0.64
110
11.5
2.881
2
0.576
1.424
0.23
107
12.0
3.230
3
0.646
2.354
0.06
4D
79.0
36.920
4
4.167
0.167
0.81
PPPC start
79.0
39.500
4
4.260
0.260
0.77
Assumptions and notes: Enrollment period from 1st dose to last dose; uniform rate of enrollment over that period; instantaneous treatment effect (ie, no treatment lag); instantaneous reports of deaths; 100% followup for mortality over defined followup period; new analysis at the time of each death.
OCR for page 256
Review of the Fialuridine (FIAU) Clinical Trials
Table H7. Analysis at the Start R90, R91, and PPPC Bases on all Deaths Regardless of Time Since Enrollment
Enrolled
Deaths
Start
No.
No. PYs
Obs
Exp
Diff
pvalue
Follow up to start of indicated trial
Low (R89 0.10; R90 0.04; R91 0.02)
R90
12
6.6
3
0.6
2.400
0.05
R91
51
49.0
4
3.1
0.900
0.75
PPPC
79
113.1
9
5.7
3.300
0.25
Intermediate (R89 0.15; R90 0.08; R91 0.03)
R90
12
6.6
3
1.0
2.000
0.16
R91
51
49.0
4
5.3
1.300
0.45
PPPC
79
113.1
9
10.0
1.000
0.66
High (R89 0.20; R90 0.12; R91 0.04)
R90
12
6.6
3
1.3
1.700
0.28
R91
51
49.0
4
7.4
3.400
0.13
PPPC
79
113.1
9
14.2
5.200
0.11
Assumptions and notes: Enrollment period from 1st dose to last dose; uniform rate of enrollment over that period; instantaneous treatment effect (ie, no treatment lag); instantaneous reports of deaths; 100% followup for mortality over defined followup period; new analysis just prior to start of a new trial. PY = personyears.