`RESEARCH
`
`APPLICA TION NUMBER.-
`
`22-341
`
`SUMMARY REVIEW
`
`
`
`Summary Basis for Regulatory Action
`
`
`January 25, 2010
`Curtis J. Rosebraugh, MD, MPH
`Director, Office of Dru Evaluation II
`
`Summar Review
`
`
`
`NDA 22-341
`NDA/BLA #
`
`SUI I #
`
`| NOVO Nordisk
`A licant Name
`Victoza
`Proprietary /
`Liraglutide (rDNA origin) Injection
`Established
`
`USA_N_) Names
`Injectable solution (6mg/mL)
`Dosage Forms /
`
`Strengt_h
`Proposed
`Indication s
`
`Introduction
`
`This review will be a brief summary of the basis for the regulatory action regarding liraglutide.
`Please refer to the reviews in the action package for a more detailed discussion. Liraglutide is
`a glucagon-like peptide-1 (GLP-l) analogue. GLP-1 is an intestinal peptide released in
`response to food ingestion that has an enhancing effect on insulin secretion when serum
`glucose is elevated and also has an inhibitory effect on glucagon (thereby inhibiting hepatic
`glucose synthesis) as well as slowing gastric emptying. GLP—l has minimal, if any, effect on
`insulin secretion when glucose is normal or low and therefore GLP—l analogues, by
`themselves, have less hypoglycemia as compared to some of the other agents used to treat
`diabetes. Intrinsic GLP-1 is degraded in minutes by dipeptidyl peptidase IV (DPP-4) which
`limits its clinical use; however, the analogues have prolonged pharmacokinetic profiles which
`allow a practical dosing interval.
`
`The Agency has recently approved one other GLP-l analogue, Byetta (exenatide) which is
`. administered twice—daily as a subcutaneous injection. In addition, several others are in various
`stages of development.
`
`As an overview, the efficacy of this drug is not in question. However, there are preclinical and
`clinical safety concerns that have led to differing opinions among the reviewers within the
`division as to whether liraglutide should be approved for marketing.
`
`Preclinical rodent studies demonstrated C-cell hyperplasia (considered a pre-neoplastic lesion
`for medullary thyroid cancer in rodents) and C—cell tumor findings in two different rodent
`species (both sexes) at clinically relevant doses. Deciding on a course of action in regard to
`this finding is new territory for the agency. As we have yet to encounter this, we have not
`determined what this finding may mean in regard to human use and the concern is that it may
`be an indication that use of this drug will plaCe humans at risk for medullary thyroid cancer
`
`
`
`(MTC). MTC is a very rare tumor in humans, with about 600 cases a year. Therefore, the
`question in regard to this preclinical finding, while probably only applying to a very small
`population, is what is the strength of uncertainty and whether or not the clinical benefit/utility
`would justify marketing in the face of uncertainty. If we were to allow marketing, this then
`brings into question what type of monitoring would we consider (if any) for What is a very rare
`event in humans. It should also be noted that the decision made in this regard does not affect
`just Iiraglutide, as Dr. Bruno-Davis has noted that data under review from other GLP—1
`receptor agonists with longer half-lives as well as sustained-release formulations of short-
`acting analogues suggest that they all have this effect in rodents and that it is probably related
`to persistent receptor activation. It is interesting to note that a sustained-release formulation
`for exenatide, also seems to exhibit this findings while the immediate—release form did not,
`although as I will discuss below, the preclinical studies for the immediate-release form did not
`reflect the frequent dosing interval used in humans.
`
`While the pre-clinical findings are the main concern with this application, there are also
`cardiovascular and pancreatic clinical issues that need consideration.
`
`Regarding cardiovascular issues, control of hyperglycemia by hypoglycemic drugs has
`consistently demonstrated benefits in microvascular outcomes (retinopathy, neuropathy, renal
`function) but not so for macrovascular events (stroke, myocardial infarct). This is not a new
`finding, as sulfonylurea drugs have carried labeling indicating that they may increase
`cardiovascular mortality up to 2.5 times that of patients treated with diet alone. However, in
`the last two to three years, there have been increasing concerns that other anti—diabetic drugs
`may also increase cardiovascular events. This has led to debate regarding the adequacy of
`cardiovascular risk assessment during development programs. This is important as
`cardiovascular disease is very common in the general population and patients with diabetes
`have an additional 2 to 4 times increase risk compared to matched non—diabetic populations.
`Therefore, from a population health standpoint, if a drug increases cardiovascular risks it
`would affect a very large number of patients. These issues were discussed at an Advisory
`Committee meeting in July of 2008, where the panel recommended that glycemic control
`agents for type 2 diabetes coming before the agency should at a minimum have some type of
`screening pre-approval cardiovascular assessment, with further, definitive, post—approval
`testing. After much internal deliberation and consideration of the recommendations we
`received from AC panel members, we issued a final guidance that incorporated
`recommendations from that meeting. This guidance, in accord with the recommendations we
`received, allows for two ‘step-wise’ assessments of potential cardiovascular risk during drug
`development. Step-one occurs during the development program before marketing, and
`requires making a determination that the investigational agent has an upper bound of a two-
`sided 95 percent confidence interval for the estimated risk ratio of less than/equal to 1.8
`compared to a control group (with a point estimate near unity). This would assure that at a
`minimum, the drug does not double the risk of cardiovascular disease. Demonstration that less
`than a 1.8 increase exists allows marketing while a longer and larger outcome study is
`conducted. The concept is that any further or more definitive pre—approval testing would be
`too burdensome to drug development, but this level of definition described above would be
`feasible/practical and would provide some assurances while further testing was underway.
`Further testing would be accomplished by a larger outcome study that must demonstrate that
`
`
`
`the investigational agent has an upper bound of a two-sided 95 percent confidence interval for
`the estimated risk ratio of less than/equal to 1.3 (rule out a 30% increase, smallest amount of
`difference felt to be generally practical) compared to a control group in order for continued
`marketing to occur.
`
`These principles incorporate recommendations from the advisory committee. The details of
`this approach are outlined in the guidance], but of relevance is that at the time of issuance of
`the guidance, three NDAs were in review. We concluded that recommendations should apply
`to all ongoing programs including those with applications pending with the agency at the time
`of guidance issuance. Although not totally in alignment with the guidance, liraglutide as well
`as saxagliptin seemed to, imspirit, fulfill ‘step-one’ and both were presented at a subsequent
`advisory committee meeting for discussion. The majority of the panel at that meeting voted
`that liraglutide (and saxagliptin) had fulfilled step-one requirements which would allow for
`marketing while awaiting the results of a definitive study.
`7
`
`Pancreatitis has been identified in post-marketing reporting with the use of incretin-based
`therapies. We have received reports for both exenatide (Byetta) and sitagliptin (Januvia—a
`dipeptidyl-peptidase IV inhibitor) and these reports also included cases of
`hemorrhagic/necrotizing pancreatitis. While the preclinical animal studies and pre-marketing
`clinical development program for exenatide and sitagliptin did not detect a signal, a recent
`publication2 in a transgenic rat model that expresses human islet amyloid polypeptide (IAPP or
`amylin) did note that rats exposed to sitagliptin increased pancreatic ductal cell turnover,
`demonstrated metaplasia of these cells and one animal had pancreatitis (hemorrhagic). While
`further exploration of these findings is necessary, this does provide a possible mechanistic
`hypothesis for pancreatitis in regard to drugs that exert there effects through the incretin
`system. The published report referred to above along with the post—market reports gives us
`great concern and will lead us to have further studies in animal models that are a closer
`approximation to the disease state of Type 2 diabetes.
`
`The preclinical evaluation for liraglutide did note increased pancreatic organ weight, but
`treatment-related microscopic pathology, overt pancreatitis or pancreatic cancer was not
`identified. Other studies conducted in a variety of animal models that give a closer
`approximation to Type 2 diabetes (insulin deficiency but not resistance as noted in Dr. Parks
`memo), did not reveal any serious gross pancreatic pathology although none of these models
`display the complete clinical presentation of diabetes and were not performed as toxicology
`studies so they did not include careful histopathology evaluations. However, there were
`several cases of pancreatitis in subjects during the liraglutide clinical development program,
`with a greater number associated with the use of liraglutide than controls, even after correcting
`for exposure. This adds to the body of evidence that is accumulating that incretin-based
`therapies may have some type of detrimental effect in the pancreas.
`
`1 Diabetes Mellitus-Evaluating Cardiovascular Risk in New Antidiabetic Therapies to Treat Type 2 Diabetes,
`December 2008, Clinical/Medical.
`2 Matveyenko AV, Dry S, Cox HI, Beneficial endocrine but adverse exocrine effects of sitagliptin in the human
`islet amyloid polypeptide transgenic rat model of type 2 diabetes: interactions with metfonnin. Diabetes. 2009
`Jul; 58(7):1604-15
`
`
`
`I will discuss these issues as well as provide an overview of the efficacy findings below.
`
`Efficacy
`
`This has'been thoroughly discussed in Drs. Derr, Yanoff, Joffe and Parks reviews and I agree
`with their conclusions. The following table from Dr. Parks review (Page 17), summarizes the
`important randomized trials.
`
`Table 7.1 Summary of Pivotal Phase 3 Studies
`Mean
`Mean Baseline
`Study #
`Treatment Groups
`- Background
`Duration of
`HbAlc
`Therapy
`
`Diabetes (yrs)
`_.
`
`Study 1573
`Lira 1.2 mg
`Diet and exercrse
`8.2
`5.4
`Lira 1.8 mg
`Glimepiride 8 mg
`
`
`Study 1572
`Lira 06 mg + met 2g
`Metformin
`8.4
`7.4
`L1ra12 mg + met 2g
`Lira 1.8 mg + met 2g
`Metformin 2g
`Glimepiride 4mg + metformin 2g
`
`.
`
`Study 1436
`
`Glimepiride
`
`Lira 0.6 mg + glim 4mg
`Lira 1.2 mg + glim 4mg
`Lira 1.8 mg + glim 4mg
`Glimepiride 4 mg
`Rosiglitazone 4 mg + glim 4mg
`
`
`Study 1574
`Lira 1.2 mg + met 2g + rest 8 mg
`Metformm +
`8.5
`9.0
`-
`Lira 1. 8 mg + met 2g + rosi 8 mg
`rosiglitazone
`Met-formin 2g + rosi 8 mg
`
`8.4
`
`7.9
`
`Study 1697
`
`Lira 1.8 mg + glim 4 mg +- met 2g
`Glim 4 mg + met 2 g
`Insulin glargine + glim 4 mg + met 2g
`
`Metformin +
`glimepiride
`
`8.3
`
`9.4
`
`The primary endpoint was change from Baseline of HbAlc either after 52 weeks (Study 1573)
`or 26 weeks (remaining studies). The review team concluded that the 0.6 mg dose
`demonstrated minimal efficacy compared to the 1.2 mg and 1.8 mg doses. The following table
`from Dr. Joffe’s review (Page 17) summarizes the findings from these trials.
`
`
`
`Table 3. Change from baseline in HbAlc (%)
`intent-to-treat 0 n ulation with last-0bservation—earried-forward
`
`N BaselineiSD
`
`Ad'usted
`J
`
`mean
`
`changeiSE
`
`Stud 1573 —52 weeks
`
`Change with lira relative
`to chan_e with lacebo
`
`Change with lira relative to
`chane with com . arator
`
`95% CI Monothera
`
`Mean difference - Mean difference
`
`95% CI
`
`p-value
`
`
`
`
`
`-0.8d:0.1
`8.2i1.1
`236
`' 'ra 1.2 mg
`-1.1d:0.1
`8.2i1.1
`234
`a 1.8 mg
`—0.5¢0.1
`8.2i1.1
`241
`Uiime- 8 m_
`Add-0n to metformin Stud 1572 —- 26 weeks
`
`-0.3 (-0.5, —0.1)
`—0.6 (-0.8, -0.4)
`N/A
`
`
`
`‘
`
`0.001
`<0.0001
`
`Lira 0.6 mg
`Lira 1.2 mg
`Lira 1.8 mg
`Placebo
`
`-0.7i0.1
`-1.0i0.1
`-1.0:t0.1
`0.1:|:O. 1
`
`-0.8 (-1.0, -O.6)
`—1.1 (-l.3, —0.9)
`-1.1 (-1.3, -0.9)
`
`<0.0001
`<0.0001
`<0.0001
`
`+0.3 (0.1, 0.5)
`0.0 (-0.2, 0.2)
`0.0 (-0.2, 0.2)
`
`<0.001
`0.88
`0.86
`
`
`
`
`
`—1.0i0.l
`Glimep 4 mg 234
`Add-on to _lime iride Stud 1436 — 26 weeks
`Lira 0.6 mg
`Lira 1.2 mg
`Lira 1.8 mg
`Placebo
`
`—0.8 (-1.1, -0.6)
`—1.3(-1.5, —1.1)
`~1.4 (-1.6, -1.1)
`
`-0.2 (-0.4, 0.0)
`-0.6 (-0.8, -0.5)
`—0.7 (—0.9, —0.5)
`.
`
`Rosi 4 mg
`
`Add-on to metformin+rosilitazone Stud 1574 — 26 weeks
`Lira 1.2 mg
`174
`8.5il.2
`-1.5:t0.l
`-0.9 (-1.1, -0.8)
`<0.0001
`Lira 1.8 mg
`177
`8.6i1.2
`-1.5:|:0.1
`-0.9 (—1.1, -0.8)
`<0.0001
`Placebo
`167
`8.4:|:1 .2
`-0.5i0.l
`
`N/A
`
`Add-0n to metformin+_lime iride Stud 1697 — 26 weeks
`Lira 1.8 mg
`224
`8.3i0.9
`-1.3i0.1
`-1.1 (-1.3, —0.9)
`Dlacebo
`110
`8.3i0.9
`—0.2i0.1
`
`r_ine
`
`225
`
`8.2d:0.9
`
`—1.1i0.l
`
`<0.0001
`
`—0.2 (—0.4, -0.1)
`
`<0.01
`
`I would like to make a few observations regarding these results. Study 1573 was the only
`monotherapy evaluation of liraglutide. More patients in the glimepiride group (10.1%)
`withdrew due to reasons of ineffective therapy compared to the liraglutide groups (3.6 and
`6%). As noted above, there was a dose-ordered increase in LS Mean Change from Baseline
`for HbAlc in the liraglutide groups that were substantially greater than the glimepiride 8—mg
`group.
`It would appear that liraglutide had greater reductions in HbAlc than the comparator
`(and less withdrawals due to ineffectiveness) in a ‘fair fight’ as the maximal labeled dose of
`glimepiride was used.
`
`Study 1572, which added glimepiride 4—mg and various doses of liraglutide onto metforrnin
`therapy demonstrated that comparisons between the glimepiride vs. liraglutide had non—
`inferiority (as opposed to superiority). This would indicate that over a 26 week period,
`liraglutide would not offer any difference in HbAlc reduction over glimepiride as add—on
`therapy to metformin. This should be viewed with some caution, as this study did not include
`glimepiride using an 8-mg dose. So it is unknown if glimepiride may have actually had
`superior results. The results of this study seem to somewhat contradict those of Study 1573, or
`at least demonstrate that there may be a difference in efficacy between these two drugs
`dependant upon whether they are used as monotherapy or are added on to another therapy.
`
`I Study 1436 demonstrated that liraglutide added on to glimepiride had additional benefits
`compared to add-on rosiglitazone therapy. However, rosiglitazone was given at only half the
`
`
`
`maximal approved dose, so these results may be somewhat misleading and would need to be
`supported by trials in which rosiglitazone is given at its maximal dose.
`
`Study 1697 offered an interesting insight into what may happen if patients were begun on
`insulin therapy instead of a GLP-l inhibitor. While it must be viewed with caution as the
`insulin titration may not have been as aggressive as what may occur in practice, it did appear
`that liraglutide at least held its own with an additional benefit of weight loss (-1.81 kg)
`compared to weight gain (+1.62) as noted in the insulin group.
`I would also point out that
`aggressive insulin titration may result in more hypoglycemia, which may be somewhat less of
`a concern with liraglutide therapy (although hypoglycemia is still a concern when incretin
`therapy is used in conjunction with insulin secretagogues).
`
`Dr. Parks has observed how the results noted above compare to other placebo-subtracted
`HbAl c changes seen with recent approvals in other trials. These are not head-to—head
`comparisons, but does give a Sense of how liraglutide might ‘stack—up’ with other treatments
`and are summarized in the table below.
`
`
`
`
`
`
`
`
`
`
`
`
`The clinical trials demonstrate that liraglutide use provides an important reduction in HbAlc,
`and for the most part, when added to already established therapy (usually metformin), has
`effects that are either the same (or better?) than other diabetic agents (at least the ones tested
`above). The weight loss associated with use and less profound effects on hypoglycemia
`(probable) are important considerations in therapy. This of course wOuld need to be weighed
`against the inconvenience of the drug only being available as an injectable and the safety
`considerations discussed below.
`
`Safety
`
`Drs. Parola and Davis-Bruno have recommended not approving liraglutide because of the
`findings of thyroid C-cell hyperplasia, adenoma and carcinoma in rats and mice. These
`findings seem to be occurring through a non-genotoxic mechanism. Their evaluation
`concluded that the human relevance of these tumors is unknown, but that there is not a
`mechanistic explanation of this phenomenon that would preclude its importance to humans.
`Dr. Paul Brown agrees with the evaluation of findings from Drs. Parola and Davis—Bruno, but
`gives options for possible actions including not approving until further pre-clinical study has
`occurred, or, if clinical benefit is considered great enough, to allow marketing with
`postmarketing requirements for further animal testing to try to better define a mechanism.
`
`While liraglutide has been negative in a series of genetic toxicity studies, it has demonstrated
`dose-related carcinogenic potential in both genders of rats and mice in life-time treatment
`carcinogenic studies occurring at clinically relevant exposures. Drs. Parola and Davis-Bruno
`
`
`
`are also concerned because in shorter duration studies, rodents have demonstrated thyroid C—
`cell focial hyperplasia which is felt to be a marker for possible evolution into C-cell carcinoma
`in rodents. The ultimate concern is that these findings in rodent may be a marker for risk in
`humans for the formation of MTC. Noted during the life-time study was C—cell adenomas
`(non—malignant) in mice and rats at 10x exposures (both genders) and 0.5-2x exposures
`(depending on gender), respectively. Also, excess carcinomas were noted in female mice (but
`not male) and male rats (but not female) at 8x and 45x exposures, respectively. The
`carcinomas were noted after 60-70% life exposure, depending on species, although the
`malignancies were not a cause of death. The two tables below from Dr. Parks review (pages 9,
`11) summarize the carcinogenicity studies.
`
`Table 4.3. Th roidHisto1 atholo Results'in Rat Carcino
`— N=50
`N=50
`N=50
`N=50
`
`enici Stud
`
`N=49*
`
`N=49*
`
`N=50
`
`
`
`2
`
`1
`
`1
`
`
`
`Follicular Cell
`
`Follicular Cell
`Carcinoma
`
`2
`8
`6
`4
`
`0
`1
`1
`1
`
`l
`2
`0
`0
`
`
`
`l
`2
`2
`1
`
`3
`9
`4
`
`
`
`2
`14
`6
`5
`
`O
`3
`2
`2
`
`2
`3
`0
`1
`
`0
`2
`4
`2
`
`0
`2
`l
`O
`
`O
`2
`O
`1
`
`(malignant)
`
`Focal C-cell
`
`
`hyperplasia
`
`
`0
`Minimal
`Mild
`7
`
`Moderate
`3
`Marked
`-
`
`
`
`
`27
`14 (28%)
`20
`..
`Ttl incidence
`11 (22%)
`
`
`(29%) r 5%) (40%) (48%) (29%)
`
`
`
`
`
`
`
`
`
`Diffuse C-cell
`
`hyperplasia
`
`
`
`Minimal
`
`Mild
`
`Moderate
`
`
`Marked
` Ttl incidence
`
`
`
`
`
`3(6%)
`
`602%)
`
`3(6%)
`
`704%)
`
`602%)
`
`8(16%)
`
`3(6%)
`
`3(6%)
`
`C-cell adenoma
`(benign)
`
`21
`(42%)
`
`23
`(46%)
`
`5
`
`(27%)
`
`(33%)
`
`28
`(56%)
`
`
`
`0cell carcinoma
`(malignant)
`
`
`
`
`
`7 (14%)
`
`
`
`
`=low dose (0 075 mg/kg/d); ID= intermediate dose (0.25 mg/kg/d); HD= high dose (0.75 mg/kg/d)
`*Note that some thyroid glands were autolysed in animals found dead and no histopath could be performed. No abnormal gross pathology
`were listed for these animals.
`
`
`
`
` Table 4.6 Thyroid Histo - athology Results in Mouse Carcinggenicity StudL
`
`
`,
`p
`..
`_; 25315163,? '
`-
`_
`.
`.
`'
`'.
`'-FemaleS--
`'.
`
`
`1
`Dose (mg/kg/day)
`0
`0.03
`0.2
`l
`3
`0
`0.03
`0.2
`Human exposure multiplea
`-
`0
`l 8
`10.0
`45.0
`-
`0.2
`1.8
`10.0
`45.0
`
`
`79
`66
`65
`67
`79
`75
`66
`67
`66
`76
`
`
`
`
`
`
`
`
`.
`b
`30
`1
`ll
`7
`i0
`22
`
`Focal hyperplasm (rare)
`0
`(2%)
`(16%)
`(38%)
`(10%)
`(15%)
`0
`(29%)
`
`
`
`
`c
`9
`15
`4
`IS
`
`
`C-cell adenoma (rare)
`0
`0
`(13‘, 0)
`(19%)
`0
`(600)
`0
`(20%)
`
`
`
`
`
`C-cell carcinoma (rare)c
`0
`0
`0
`0
`0
`0
`0
`2 (3%)
`
`
`
`C—cell adenoma or
`0
`0
`9
`i5
`0
`:5
`0
`a7
`
`
`
`
`
`
`carcinomafireL
`(13%)
`“9%)
`(6" o) 22%)
`
`
`
`a=based on area under the time-concentration curve relative to the 1.8 mg dose
`
`
`
`b=Diffi1se C-cell hyperplasia cannot be adequately assessed without specialized staining
`c=tumor considered common or rare based on incidence in historical control groups of >1% or <1%, respectively
`
`
`
`statistically significant differences fi'om control identified in bold/red font
`
`
`
`0
`0
`0
`0
`
`0
`0
`0
`0
`
`Mechanistic studies to date have not been able to determine that this finding is limited to
`rodents only. Diffuse and focal hyperplasia and adenomas are common findings in aging rats
`(but not mice), however carcinoma is a rare finding (<l%). As stated above, the carcinomas
`were not discovered until after 60-70% of the rodent life-span (depending on species) but C-
`cell hyperplasia was noted in mice exposed to 88x exposure compared to humans within 4
`weeks, which would correspond to 3-4% of their lifespan. However, data also indicates that
`the hyperplasia mostly reverses after a 15 week recover period (one animal had not reversed
`yet) as is demonstrated in the table below.
`
`Table 5‘.
`
`\FZMSEB. Incidence focal C-t'ell ll}'|)l‘)'|)l1’l5lfl in mice dosed with lirnglutidc for '0
`weeks followed by :1 G 1i'ttkt9+61\'k)v)t‘ 15 week (9+15wk1 recovery period (30-
`54:3y0upfiime. some cunnbinurl}.
`Tim:- p-oinl
`:\‘n. of animals vu'lb Inch! C-cl-ll hyperplmlu 1' turn! numbm' in
`:10in
`[Rum I’llu'kufiifly
`
`
`
`
`Il-I‘i‘lnli’d hone-5: is $wa in Mn! (lean Imx‘qflaiiituntr‘fl week-scream; stuns Itu‘l‘sibilily arm r. :.-'-3
`Therm:
`[5 -.x mks ohm-fiery.
`
`Monkey studies have not shown proliferative thyroid C-cell lesions following liraglutide
`treatment up to 20 months (5% of their lifespan) at > 60x human exposure. This must be
`viewed with some caution as GLP-l receptors have also not been demonstrated in monkey
`thyroid tissue and as Dr. Davis-Bruno observes, there were a limited number of animals
`(n=40), limited life—time exposure and immunogenic response in monkeys that may have
`neutralized the effects in monkeys. Dr. Parks notes that the immunogenic response was only
`noted in 3 animals receiving the highest dose and was expressed only after 52 weeks of
`exposure. To complete the picture of GLP~1 receptors in thyroid tissue, Dr. Parola has done
`an extensive search and has located only one paper3 where autoradiography detected GLP—l
`binding in 12/12 normal thyroid samples from rats, 3/5 normal thyroid samples from mice, and
`
`
`3 Komer M, et al. GLP-1 receptor expression in human tumors and human normal tissues: potential for in vivo
`targeting. Jn Nucl Med. 2007; 48:736—43.
`
`
`
`1/18 normal human thyroid samples from surgical resection, with 5/18 surgical resection
`samples from subjects with MTC demonstrating binding. The sponsor submitted a summary
`of an autoradiographic ligand binding study during the review cycle where they were unable to
`duplicate the above results in humans (while they could in rats) but has not submitted the study
`for review. I think it is hard to draw final conclusions on what the true binding in normal
`human thyroid tissue may be without further study, but this may be an indication that there are
`GLP—l receptors in the thyroid of humans, although it does not locate whether the binding
`occurs in follicular cells or C-cells.
`
`For clinical considerations, calcitonin is synthesized and secreted from C-cells and is used
`clinically as a screening tool and prognostic biomarker for MTC. Therefore, a great deal of the
`clinical safety analysis for this application is focused on measuring calcitonin and what trends
`the levels may have exhibited during the clinical trials. Normal levels for females and males
`are 5.0 ng/L and 8.4 mg/L respectively and levels of 50 ng/L or greater are considered
`clinically important warranting further evaluation. The sponsor has provided data with up to
`600 liraglutide-treated subjects after 2 years of treatment. Dr. Joffe has a very thorough and
`complete outline regarding the limitations of the data, including limitations on blinding for
`extension data beyond six months and such, but I would agree with him that calcitonin is an
`objective measure and does provide insight into serum calcitonin trends during treatment.
`I
`would note that the safety reviews of Drs. Mahoney, Joffe and Parks have very thorough
`analyses and discussion that go into great detail regarding whether or not there are trends that
`may indicate dose-related shifts of calcitonin'levels, Whether there are imbalances of the
`percentage of liraglutide exposed subjects with any upward shifts compared to the overall
`population, and other forms of exploratory analyses along with their interpretations of what the
`results demonstrated. However, to put this in a broader context, all these analysis involve very
`small changes in serum calcitonin, from below the level of quantification (LOQ), which is 0.7
`ng/L, to around 1 ng/L.
`I think, considering that levels of 50 ng/L'are considered the point at
`which it is important to consider cancer, and that there is usually a lot of variability involved in
`measurements that are around the LOQ, trying to draw conclusion regarding trends of values
`that are less than one is giving more credibility to an imprecise measure than is warranted. Dr.
`Joffe has the table below that gives a summary of the six month controlled data (page 38).
`
`
`Table 14 Re - eated measures anal s1s for calcrtonin for the five ma or base 3 trials
`
`
`
`
`
`LS mean 95% CI
`
`
`
`
`
`
`0.78 (0.72, 0.84)
`Liraglutide 0.6 mg
`0.78 (0.73, 0.83)
`Liraglutide 1.2 mg
`
`0.76 (0.72, 0.81)
`Liraglutide 1.8 mg
`
`0.70 (0.66, 0.74)
`Active comparator
`
`0.67, (0.63, 0.73)
`Placebo
`
`
`
`
`
`
`
`
`
`LS mean 95% CI
`
`0.96 (0.90, 1.04)
`0.99 (0.94, 1.05)
`1.01 (0.95, 1.06)
`0.97 (0.91, 1.02)
`0.89 (0.83, 0.95)
`
`
`Considering that the LOQ is 0.7 ng/L, and that the upper level of normal is 5 for females and
`8.4 for males, this table indicates to me that there is not an important effect of liraglutide on
`serum calcitonin. Below are several other figures from Dr. Joffe’s review that demonstrate
`data from longer periods of evaluation during open—label extension phases of the clinical trials.
`
`
`
`Figure 2. Geometric mean calcitonin values over 2 years in the monotherapy trial (Week
`0-52 is blinded; Week 52-104" is the open-label, voluntary extension)
`Zr-Liraglutide 1.8 mg
`--Liraglutide 1.2 mg
`"Ethlimepiride 8 mg
`
`10
`
`Upper normal range males (8.4 ng/L)
`____________________________________________________________________________
`
`9
`j 8
`Bu
`5 7
`Upper normal range females (5.0 ng/L)
`.E
`6
`c 5 """"""""""""""""""""""""""""""""""""""""""""
`.43 4u
`E 3
`U 2
`
`-
`.
`- a»
`”tn-1.:- M . 1%.;
`. "we:
`
`5;“
`1 kW
`;'__rj:‘__:“n,‘;;r=b—y§€fim.-WP%4TE"W%W%}TN '
`-’--:---'-” 4: "1
`04—‘4—r—r—r—fifl—l—l—y——l—1——I
`.0
`8
`16
`24
`32
`4Q
`48
`56
`64
`72
`80
`88
`96
`104
`
`Time (weeks)
`
`Figure 3. Geometric mean calcitonin values over 2 years in the add—on to metformin trial
`(Week 0-26 is blinded; Week 26—104 is the open-label, voluntary extension)
`+Liraglutide 0.6 mg
`~-:—.-_-.—. Liraglutide 1.2 mg
`~-.sI.-;.~ Liraglutide 1.8 mg
`~®~Glimepiride
`«em Placebo
`
`l—L
`
`~i
`
`Upper normal range males (8.4 ng/L)
`
`
`_l: r
`“I
`'
`I
`
`F
`
`OHNw-hU'IOWVGMOO
`
` Calcitonin(ng/L)
`
`16
`
`24
`
`32
`
`40
`
`48
`
`56
`
`64
`
`72
`
`8O
`
`88
`
`96
`
`104
`
`Time (weeks)
`
`Figure 4. Geometric mean calcitonin values over 1 year in the obesity trial (at Week 52,
`patients on liraglutide and placebo were switched to liraglutide 2.4 mg and subsequently
`uptitrated to 3.0 mg)
`
`10
`
`
`
`10
`
`9
`
`8
`
`Upper normal range males (8.4 ng/L)
`
`_
`
`_ _
`
`
`
`
`
`Placebo group and all
`liraglutide groups
`switch to optimal dose
`
`,
`
`
`
`o
`
`8
`
`
`
`20 32 44 52 64 76 88100104
`
`Weeks
`
`+Liraglutide 1.2 mg
`.
`Liraglutide 1.8 mg
`.
`.
`—<‘>— Liraglutide 2.4 mg
`Upper normal range females (5.0 ng/L)
`5 ----------------------------9— Liraglutide 3.0 mg
`
`wycmOrlistat 120 mg TID
`”315‘“ Placebo
`
`6
`
`3 7
`3
`5
`c
`'5
`:03
`U
`
`Tu
`U
`
`OHNUJ-fi
`
`The results presented in these tables give me comfort that liraglutide is not causing a mean
`change in calcitonin levels compared to other drugs or placebo, and is not even approaching
`the upper normal range in females or males, at least up to a two year time interval. Dr. Joffe
`details 17 subjects who have at least one.
`I refer the reader to his very thorough analysis, but I
`agree with his conclusion that there is not an imbalance in treatment—emergent serum
`calcitonin 220 ng/L between the liraglutide treated subjects vs. comparators. Dr. Joffe notes
`that the two thryoidologists sitting on the Advisory Committee panel had a split vote regarding
`approval based on the thyroid findings. I note that the thyroidologist voting no during his
`discussion stated that if he had data from a 6-12 month extension study documenting that
`calcitonin levels did not continually raise, he would feel assuaged (page 219, AC transcript).
`The tables above would seem to fulfill that criterion.
`
`Dr. Joffe has a very thorough review of the likelihood that additional clinical data would
`feasibly define human risk. As the table below from his review (page 59) demonstrates, given
`the rarity of the tumor, there would have to be at a minimum 100-fold increase in the incidence
`of the cancer for detection. This seems highly unlikely (even the rodent models did not have
`carcinomas above baseline rates at doses approximating human exposures) and also indicates
`that this is not likely a question to be answered by a clinical trial.
`
`
`
`
`
`
`
`
`
`Table 22. Sample sizes needed for a single-arm trial to detect 2—fold to 100-fold
`
`increases over the background rate in the risk for non-familial medullary thyroid
`carcinoma from Drs. Derr and Sahlroot, FDA biostatisticians
`
`Power
`Increase in risk for medullary thyroid carcinoma
`
`3-year treatment period
`
`
`80%
`229,783
`1,888,050
`
`
`ll
`
`
`
`90%
`
`5-year treatment period
`80%
`90%
`
`
`2,578,783
`
`
`
`
`
`1,132,830
`
`
`1,547,270
`
`
`
`
`
`
`
`10-year treatment period
`68,935
`566,415
`80%
`83,515
`773,635
`90%
`Power calculations from' StatXact, l-sided al ha 005
`
`
`
`
`
`278,383
`
`110,850
`
`137,870
`167,030
`
`
`
`40,517
`
`18,710
`24,310
`
`9,355
`12,155
`
`37,430
`66,510
`
`18,715
`33,255
`
`In regards to the cardiovascular safety evaluation, please see Dr. Parks review for a
`comprehensive discussion with which I am in agreement. The filing of this application
`predated Agency guidance regarding cardiovascular safety evaluation for drugs used in
`glycemic control in diabetes. As such, this program did not have pre-specified definitions or
`prospective adjudication of major cardiovascular endpoints and any evaluation was
`retrospective in nature. Therefore the cardiovascular event data were evaluated in many
`different ways. I believe the most pertinent aspect of this is that for the most part, no matter
`how the data are ‘sliced and diced’, the point estimate for cardiovascular events is less than
`one, and when all categories are considered such that there are enough events to have any
`meaning, the upper bound of the confidence interval is less than the 1.8 goal-post (bearing in
`mind all the caveats inherent in any type of unplanned, retrospective analysis on an endpoint
`that was not originally identified as something of interest and where only a small number of
`events occurred). This does give us some reassurance that liraglutide will not have a negative
`cardiovascular impact and while not a perfect analysis, does fulfill the, spirit of our guidance
`and the Advisory Committee voting reflected this View point as well. I believe that due to this
`being an unplanned analysis, retrospective in nature, and with limited events, labeling should
`not include this analysis and should only have the standard labeling that we use for diabetic
`I
`agents noting no conclusive evidence of cardiovascular benefit with any anti-diabetic drug.
`believe to do otherwise would be misleading, would misrepresent our degree of comfort with
`the data, and would create an unfair playing field for other agents.
`
`I do note that the Advisory Committee Panel, with a clear majority voting that there was
`enough cardiovascular data to allow marketing, was not as overwhelming positive for this drug
`as it was for saxagliptin that was discussed on the previous day ofthe 2-day meeting.
`Comments at the time of those voting not to approve reflect the limited number of events upon
`which to draw conclusions. I find this somewhat curious, as this program had approximately
`the same number of events as saxagliptin, and one could speculate that having more time to
`consider things had caused some members to rethink their position. In any regard, I note as
`' Dr. Parks has that some subgroup analyses resulted in a confidence interval that exceeded 1.8,
`but, I feel that the application provides the same level of confidence regarding cardiovascular
`safety as that provided by the saxagliptin analysis (which was approved) and should the