July 27, 2001
Peter Lurie, M.D., M.P.H.
Deputy Director, Public Citizen’s Health Research Group
On the Safety and Efficacy of Teriparatide
Before Endocrinologic and Metabolic Drugs Advisory Committee
Thank you for the opportunity to testify before the Committee on the subject of teriparatide (Forteo; Eli Lilly). Based on our review of the Food and Drug Administration’s (FDA’s) assessment of the clinical data as well as several published articles on teriparatide efficacy, we do not believe that a sufficient basis for approving the drug in either men or women exists at the present time.
Four phase 3 trials of teriparatide for the treatment of osteoporosis in women and men were included in the FDA’s review. Two studies were placebo-controlled, one with postmenopausal women (GHAC), and one with men (GHAJ). Two studies were active-controlled, both with postmenopausal women: one (GHAF) with teriparatide as an add-on to hormone replacement therapy (HRT), and one (GHAH) with alendronate.
1. GHAC (1,637 postmenopausal women. Placebo, 20ug teriparatide, 40ug teriparatide)
Entry criteria included presence of one or more atraumatic vertebral fracture. The primary efficacy outcome variable was incidence of new vertebral fractures, as determined by analysis of spine radiographs at baseline and study end. Median exposure to the drug was 19 months before the trial was prematurely terminated by the sponsor due to the rat carcinogenicity data described below.
The 20ug and 40ug doses yielded relative risk reductions for new vertebral fractures of 65% and 69% respectively. The corresponding absolute risk reductions were 9.3% and 9.9% respectively. This yields an number needed to treat (NNT) to avert a new vertebral fracture of about 10 over 19 months for both doses (compared to a study of alendronate in similar patients in which the NNT was 14.3, but with almost three years of follow-up (Black, et al. Lancet 1996;348:1535-41). As the reviewer notes, “[b]ecause the majority of morphometric vertebral fractures are clinically silent, it is difficult to evaluate the overall direct clinical impact of these data taken alone.” Indeed, the “sponsor did not include an analysis of clinical [symptomatic] vertebral fractures…in this application.”
For non-vertebral fractures, the relative risk reduction in the 20ug dose was 35%, with a corresponding absolute risk reduction of 3.6%. The corresponding NNT for preventing a non-vertebral fractures is 27.8 over 19 months (for the 40ug dose, it was 25.6). The NNT for non-vertebral fractures in the alendronate study mentioned above was 35.7. Relative risk reductions over three years with currently approved “anti-resorptive” drugs (alendronate, risendronate, and raloxifene) in studies of similar populations ranged from 10%-39%, according to the reviewer.
Probably because so many vertebral fractures are asymptomatic, there was no improvement compared to placebo on a Quality of Life scale, even on two osteoporosis-related measures.
2. GHAJ (437 men with osteoporosis. Placebo, 20ug teriparatide, 40ug teriparatide)
The primary efficacy outcome in this study was lumbar spine bone mineral density (BMD) following two years of treatment. Because of premature study termination, patients received treatment with active drug or placebo for only approximately 300 days (10 months). As noted by the reviewer, “[t]he trial was not intended to detect treatment group differences in fracture rates.” While the risk of fracture is assumed to be inversely related to BMD, the reviewer noted that “the risk estimates for a given BMD T-score in men are not as well determined as in women…whatever the cause of uncertainty, the clinical impact of changes in BMD will be more difficult to judge in men, compared to women, in the absence of fracture data.” For this reason, fracture data should be required for men if a drug is to be approved.
For the primary endpoint of change in lumbar spine BMD, the 20ug dose showed an absolute increase in lumbar spine BMD of 5.4% compared to placebo. For the 20ug group, BMD was also increased compared to placebo in the femoral neck, but not in the distal radius.
Comparing BMD data for teriparatide to alendronate (historical controls) is particularly instructive, especially given the significant safety concerns we address below. The reviewer notes, “at the lumbar spine, the BMD responses to [teriparatide] at 11 months were similar to those found in response to alendronate, in a similar osteoporotic male population, at 24 months … However, at most non-vertebral skeletal sites (e.g., the trochanter, the femoral neck, the total hip, and total body) the effects of [teriparatide] 20ug/day for 11 months appear to be inferior to those that followed treatment with alendronate 10mg/day for 24 months.” In fact, according to the reviewer, “there is very little evidence that this dose [20ug] has beneficial effects at other skeletal sites, with the exception of the femoral neck. In contrast, alendronate increased BMD over placebo at the femoral neck, trochanter, total hip, and total body…” He concludes “since we have no fracture efficacy data for either drug in men, it is difficult to conclude that [teriparatide] 20ug/day offers any advantage over current therapy.”
As with the placebo-controlled study in women, there was “very little within- or between-group change from baseline, in any of the measured [quality of life] parameters,” according to the reviewer.
3. GHAF (247 postmenopausal women. 40ug teriparatide + hormone replacement therapy (HRT) vs. HRT alone
The utility of the data in this trial is limited by the inclusion of only a 40ug dose of teriparatide, a dose that the sponsor seems no longer to be pursuing. Increases in BMD in the group receiving teriparatide and HRT compared with HRT alone were found at the spine, total hip and femoral neck. However, ultradistal radius and whole body BMD in patients receiving teriparatide and HRT were found to be significantly increased over the group receiving HRT alone only in subjects who had not been treated with HRT prior to the study. There were no fracture data.
4. GHAH (146 postmenopausal women. 40ug teriparatide vs. 10mg alendronate)
Increases in BMD at the lumbar spine, total hip, and femoral neck were significantly greater in the teriparatide group compared to alendronate. At the distal radius, the BMD in the teriparatide group was significantly less than the alendronate group at the conclusion of treatment, and in fact, represented a decrease in BMD at that site. As with the GHAF study above, there was only a 40ug dose of teriparatide and no fracture data.
1. Bone tumors in rats
The rat carcinogenicity data are some of the most striking animal carcinogenicity data we have ever seen. Osteosarcomas occurred at frequencies of 0%, 5%, 35% and 52% (control, low, middle, and high dose) in males and in 0%, 7%, 20% and 38% of females. No “no-effect level” for osteosarcomas was established because tumors were present at even the lowest dose level. Osteoblastomas were also statistically significantly increased in males and in females.
These data present a compelling case for the carcinogenicity of teriparatide for several reasons:
- The increases in tumors are substantial and statistically significant
- The increases are dose-related
- No no-effect level was demonstrated
- The higher the exposure, the shorter the time to tumor initiation and death
- The increases in tumors occur in both genders
- The exposure levels are small multiples of human exposures; at 18 months the lowest dose was only approximately 1.6 times (based on area under the curve) or 5 times (based on maximum concentration) the exposure in humans
- Osteosarcomas are very rare tumors in experimental animals; the FDA safety review states that they were seen in control animals in past rat carcinogenicity studies at rates of 0.3% (1/360) in males and 0.0% (0/360) in females
- The tumors are mechanism-based; teriparatide causes increases in bone formation and thus, bone is where one would expect to see tumors
- Parathyroid hormone-induced osteosarcomas have been observed in other rodent studies; FDA’s Draft Guidance for Industry on Development of Parathyroid Hormone for the Prevention and Treatment of Osteoporosis (May 2000) mentions positive studies in two rat strains and one mouse strain.
Because the formation of osteosarcomas is mechanism-based, the absence of positive results in mutagenicity and genotoxicity tests is not relevant. A likely mechanism for carcinogenicity is drug-induced cell division, increasing the chances for generating genetic errors and then magnifying them.
Moreover, the rat study may actually have underestimated the incidence of bone tumors. Tumors were detected either by clinical detection of a bone nodule or by microscopic evaluation, but only four bones were examined routinely under the microscope. The limits of clinical detection are illustrated by one tumor that was fatal yet was only detected upon microscopic examination.
Carcinogenicity studies must be powered to detect rare events; that is why group sizes are relatively large (60 rats/sex/group) and drug exposure long (2 years). This is why negative results in shorter studies with fewer animals and limited bone histology, such as those in monkeys the sponsor cites as evidence for lack of carcinogenicity, are not considered valid. Similarly, the lack of increased rates of bone tumors in patients with hyperparathyroidism is of limited relevance, because the dosing schedule here is intermittent.
2. Renal toxicity in humans and animals
Renal toxicity which reversed upon drug discontinuation was seen in several monkey studies. There was evidence of medullary interstitial expansion associated with increased deposition of extracellular matrix on histopathology. In one of these studies, the no-effect level could not be determined but at best was the same as the 20ug human dose, based on area under the curve. It is worrisome that in study GHAC there was a statistically significant increase in median serum creatinine.
3. Cardiovascular toxicity in humans and animals
Phase I clinical studies show a consistent pattern of statistically significant, dose-related decreases in supine and standing diastolic blood pressure, with accompanying increases in heart rate, “consistent with drug-induced vasodilation and a compensatory cardiac response,” according to the pharmacology/toxicology reviewer. Similar changes in hemodynamics were apparent in rats, dogs and monkeys.
On electrocardiogram, there were decreases in RR and heart-rate-corrected QT intervals in a Phase I study. The sponsor has made a Phase IV study commitment to further evaluate these findings in osteoporosis patients, although it is unclear why this study could not already have been completed.
Given the drug’s mechanism of action, it is not surprising that there was a statistically significant increase in the percentage of patients with 1 episode of hypercalcemia on 20ug teriparatide compared to placebo 4-6 hours after the dose (11.1% vs. 1.5%). Serum ionized calcium was also increased in monkeys 4-8 hours after the dose.
We do not believe that the data presented by the company provide an adequate rationale for approving teriparatide in men. There is no evidence that the drug reduces fractures in any location. Eli Lilly only has data on BMD to support this indication. We do not believe that BMD data alone should be sufficient to support a claim of efficacy, particularly in men. Moreover the carcinogenicity data are extremely convincing and outweigh any theoretical benefit to be gained from the drug.
The situation in women is more complex, as there actually is evidence of reduced fracture risk in both vertebral and non-vertebral locations. However, the absolute fracture reductions are not large, many of the fractures are asymptomatic and there was no overall impact on the patients’ quality of life. Moreover, there already are four drugs approved by the FDA for the treatment of osteoporosis. We therefore believe that the risk-benefit assessment tips against approval of the drug in women as well. However, should the Committee vote to approve this drug, it is critical that a. the drug is restricted to use as a second-line drug, to minimize population exposure to this carcinogen; b. there be a black box warning on the carcinogenicity findings; c. patients be provided with an FDA-approved Medguide, as distinct from the FDA-approved package insert for physicians or drug industry-sponsored and often misleading patient information leaflets at pharmacies; and d. that the sponsor be required to establish a registry in order to identify patients for a case-control study of whether teriparatide is associated with osteosarcoma in humans.