Tuesday, October 26, 2010

Lorcaserin CRL interpretation

This is one MD's interpretation of Lorcaserin's Complete Response Letter from the FDA.

What do the CRL requirements for Lorcaserin mean in practice?

1. Non-clinical issues

a. “Detailed accounting of all microscopic pathology slides prepared from FEMALE rats that contributed to the mammary tumor incidence data in each update to the FDA and the final study report”

i. Requirement: account for all slides done on those tissue – exactly what it says – an accounting issue.

ii. In total, 65+65+75 = 205 FEMALE rats were given lorcaserin along with 65 untreated control FEMALE rats - see Table 6 in the section: “Genotoxicity and Carcinogenicity Assessment For Lorcaserin” (Mammary Tumors).

iii. This task should not take more than 4 months, leading to a revised report (if needed)

b. Independent pathologists/group of pathologists to re-adjudicate all mammary and lung (unclear why lung) tissues from all FEMALE rats (205)

i. I understand this to mean that since a discrepancy was found between the week-96 tumor incidences at all doses – the reason for the accounting check – these pathologists are to re-read the histological slides and give a final decision on what the true incidences are, so as to settle the discrepancy.

1. This procedure should not take more than 4 months

2. The response does not require a new study, according to advice we have received from a non-clinical safety assessment toxicologist.

ii. During the Arena Conference Call on October 25, 2010, the reason for the discrepancy was provided: The preliminary slides were reviewed on a periodic basis by one pathologist and sent to the FDA prior to the final submission. However, at the end of the study, three independent pathologists performed a peer review of the data and submitted their results, the ones that were included in the final NDA.

iii. The discrepancy between the assessments made by the different pathologists was mentioned by in the section: “Genotoxicity and Carcinogenicity Assessment For Lorcaserin” (5th paragraph; Mammary Tumors), where Dr. Alavi (FDA’s nonclinical pharmacology/toxicity presenter) makes the following statement:

“In subsequent updates and in the final study report, the incidence of adenocarcinoma in the MD and HD females was lower than that reported at week 96 (Table 7a). The incidence of adenocarcinoma increased in the controls and stayed consistent in the low dose group over the same period. The incidence of fibroadenoma increased in all dose groups from week 96 to the final study report, though the numbers notably varied in the mid- and high dose groups (Table 7b). It appears that some of the decrease in the number of adenocarcinoma after week 96 was accompanied by an increase in fibroadenoma, potentially a consequence of the sponsor/CRO reclassifying the observed tumor types.”

1. The question that begs an answer here is: Why did Dr. Alavi not know that the slides had been reviewed by a single pathologist during the study and then by three independent pathologists at the end of the study?

2. It is well known that inter-observer variation can occur in assessments of this nature. This should have been taken into account, rather than implying that the sponsor/CRO had acted improperly in respect of the discrepancy in the final numbers

c. Demonstrate that the apparent increase in aggressiveness of adenocarcinoma in rats administered lorcaserin is REASONABLY (not conclusively) irrelevant to human risk assessment.

i. Here they are asking: With regard to the re-adjudicated results, a statistical analysis must again be carried out on the incidence of fibroadenomas and adenocarcinomas

ii. This is not a new study, but a re-assessment of the existing one. Unless the accounting yields a new surprise, then a further study will not be necessary.

iii. The results will most likely be the same. If they are the same, then Arena needs to demonstrate that the adenocarcinomas seen at the very high and toxic Lorcaserin doses, and the statistically significant increases in fibroadenomas, do not portend a risk to humans. The following information is relevant here:

The incidence of malignant adenocarcinoma tumors identified in the 10 mg/kg/day female group was no different to the normally-occurring incidence of these tumors identified in the untreated control group. Furthermore, in the 30 mg/kg/day female group, a dose 24-fold greater than the anticipated human therapeutic dose, the incidence of malignant mammary tumors was no greater than the normally-occurring malignant mammary tumor incidence reported in the untreated control female rats. Only at 100 mg/kg/day was there a statistically significant increase in adenocarcinoma incidence, but this lorcaserin dose is 82-fold that intended for human use. (Note that if Arena decides to fall into the FDA ‘mechanism of action (MOA) involves prolactin’ “trap”, they will fail and spend an eternity attempting to find the answer. They should not attempt to open Pandora’s box. There are numerous drugs on the market today for which the MOA for pathology is not understood).

Although the fibroadenoma incidence in the rats was found to be statistically significant at all doses, and the practice of combining benign tumors with malignant tumors is commonly done when the cell types are the same, this does not pose a human risk for the following reasons:

1. Fibroadenoma and adenocarcinoma arise from cell types with different histogenesis. Adenocarcinomas are classified under the epithelial histotype, fibroadenomas under the epithelial-stromal histotype (Russo, Gusterson, et al. 1990 and Russo, Russo, et al. 1989).

2. Benign mammary fibroadenomas can only be transformed to malignant mammary carcinosarcoma and never to mammary adenocarcinoma (Russo, Gusterson, et al. 1990 and Russo, Russo, et al. 1989). These are distinctly different tumors.

3. The rat model of tumorigensis closely mimics human breast tumor development.

4. Fibroadenomas rarely progress to adenocarcinoma in the rat.

5. Fatalities from benign fibroadenomas do not translate as a risk to humans.

6. Fibroadenomas rarely progress to adenocarcinoma in the human female and are relatively common (London, et al. 1992).

d. Provide ADDITIONAL DATA/INFORMATION regarding the distribution of lorcaserin to the CNS in animals and human subjects that would clarify or provide a better estimate of astrocytoma exposure margins.

i. Since brain partitioning – brain-to-plasma (BPD) ratio - was not determined in humans, the FDA is concerned that estimates of safety margins based on extrapolation from monkey brain-to-plasma ratios are not reliable. This assumption was made by Dr. Alavi as outlined below:

ii. The problem here is as follows:

1. The brain-to-plasma ratio for lorcaserin is not known in humans since it is a novel new drug that has not yet been studied in this way. But we do know:

a. The brain-to-plasma ratio for lorcaserin for mice is 25 times higher in the brain vs. plasma (of note, there were no brain tumors in mice including the high dose group).

b. The brain-to-plasma ratio for lorcaserin for rats is 29 times higher in the brain vs. plasma.

c. The brain-to-plasma ratio for lorcaserin for monkeys is 10 times higher in the brain vs. plasma.

2. Regarding the reliability of extrapolating monkey brain-to-plasma ratios to human subjects:

Dr. Alavi’s assumption is: "Brain partitioning in human subjects was not determined. Thus, estimating safety margins based on assumptions of partitioning in human subjects is not entirely reliable. Assuming that the monkey best models human partitioning, the estimated safety margin to a non-tumorigenic dose in rats may range from 11x to 17x, with tumors associated with brain exposures that are 40x to 59x higher than clinical exposure. More conservatively, safety margins based on plasma drug levels, which is known for rats and humans, yields a safety margin to the non-tumorigenic dose in rats of 5x, with brain tumors occurring at doses of lorcaserin 17-fold higher than the clinical dose” from the section in the FDA briefing document: from the section “Genotoxicity and Carcinogenicity Assessment For Lorcaserin” (Abstract).

iii. The solution: An extensive review on this very issue (Shen, Artru and Adkison 2004) contradicts Dr. Alavi's opinion that estimating safety margins based on assumptions of partitioning in human subjects is not entirely reliable. Referring to the monkey model, the authors state - “In the second part of our analysis, we examined the issue of whether CSF penetration studies in animals are predictive of human data. We obtained animal and human CSF data on 27 drugs, including 13 antiepileptics, 1 psychotropic drug, 5 anesthetics or analgesics, 5 antibiotics, 1 antiretroviral, and 2 anticancer drugs, and across six animal species, including rats, dogs, rabbits, cats, guinea pigs and monkeys. As can be seen in Fig. 9, there is a reasonably good correlation for the majority of drugs in this survey."

1. In the same article the authors conclude: "Despite the complexity of CSF physiology and pharmacokinetics, CSF penetration studies in animals remain a practical option for the assessment of CNS drug delivery in early preclinical drug development"

a. Monkey brain partition studies can, therefore, be used with reasonable assurance for estimating a drug's margin of safety in the human brain.

iv. The brain-to-plasma ratio for lorcaserin for rats is 29 times higher in the brain vs. plasma and the for monkeys it is 10 times higher in the brain vs. plasma

1. The interpretation:

a. This means that there is 29x more lorcaserin in the rat brain, so the dose given to rats at the LD, MD, and HD will be much higher in the CSF and therefore more toxic.

b. In practice, this means that at any given dose, the higher the brain exposure in rats, the higher will be the estimated brain exposure in humans - From the Tables 13 and 14 in the FDA briefing in the section: “Genotoxicity and Carcinogenicity Assessment For Lorcaserin (Brain Astrocytoma)”

i. Brain exposure in the rat at 30mg/kg (brain tumors present in the rat) = 405-591 mcgm/ml.

ii. Using the brain-to-plasma ration of 10x in the monkeys (multiple the brain exposure in rats by 10).

iii. Brain exposure in humans at the 30 mg/kg (brain tumors present in the rat) = 40x-59x multiple of the clinical dose (10mg bid (twice a day).

iv. Therefore, from the Table, we can see that the margin of safety is at 11x-17x multiple of the clinical dose, which would satisfy the concern brought up in the CRL.

v. The 17x dose is when the tumors first appeared (the 30 mg/kg dose). Hence the tumorigenic dose for humans, when discussing brain/plasma ratios, would then be 40-59x multiple of the clinic dose - far above the 25-fold limit dose in the FDA guidelines.

vi. So from the data on the male rats the risk of developing statistically nonsignificant astrocytomas in the human, assuming that this can be transferred to human risk, would only arise at 40-59 x multiple of the clinical dose. The margin of safety dose would be 17x multiple and perhaps anywhere up to 39x multiple, although the precise value would have to be determined at incremental increases in the dose of lorcaserin (which is not necessary).

vii. However because Dr. Alavi used plasma levels, he concluded that the nontumorigenic dose (level where there are no tumors) gave a margin of safety at only a 5x multiple of the clinical dose. Our analysis suggests that he erred in doing this.

v. Astrocytoma was only statistically significant in male rats in the high dose group, but none of these tumors was found in the female rats.

1. The solution:

a. Arena must supply all MBTD and brain-to-plasma ratio data to FDA in a simple, easy-to-read format.

b. Given adequate resources, this administrative task should not take long.

2. Clinical issues

a. Results on the BLOOM-DM trial. There is nothing more to say on this point.

3. Labeling requirement.

a. Schedule IV: This is not a significant issue. Schedule IV does not necessarily preclude long-term use, although FDA may add restrictions to that effect.

b. They did leave the door open for removal of this label.

c. Arena must discuss with FDA any nonclinical studies conducted to address abuse concerns regarding the long-term use of lorcaserin.

In conclusion, no new studies need to be done except perhaps brain partitioning studies on humans. However, this may not be necessary if the information provided above is correct. If such studies were performed, they would not be long-term experiments and could be completed in a relatively short period, and they may not even need to be carried out under GLP conditions. However, I sense Arena already has all the data required to address this particular concern. Every other aspect of the CRL involves a REVIEW of existing information, nothing else. If everything is submitted in a timely manner, the estimated timeframe for the completion of these reviews should be less than 6 months.


London, S.J., Connolly J.L., S.J. Schnitt, and G.A. Colditz. "A Prospective Study of Benign Breast Disease and the Risk of Breast Cancer." JAMA 267 (1992): 941-4.

Russo, Jose, Barry A. Gusterson, Adrianne E. Rogers, Irma H. Russo, Seft R. Wellings, and Matthew J. Van Zwietien. "Biology of Disease: Comparative Study of Human and Rat Mammary Tumorigenesis." Laboratory Investigation, 1990: 267.

Russo, Jose, Irma H. Russo, Matthew J. van Zwieten, Adrianne E. Rogers, and Barry A. Gusterson. "Integument and Mammary Glands of Laboratory Animals." In Classification of Neoplastic and Non-neoplastic Lesions of the Rat Mammary Gland, edited by T.C. Uones, U. Mohr and R.D. Hunt, 275-304. Berlin:Springer-Verlag, 1989.

Shen, Danny D., Alan A. Artru, and Kimberly A. Adkison. "Principles and Applicability of CSF Sampling for the Assessment of CNS Drug Delivery and Pharmacodynamics." Advanced Drug Delivery Reviews, 2004: 1825-1857.

Daniel P. Lopez, M.D., F.A.C.O.G.

Diplomate American Board of Obstetrics and Gynecology

BORG Member

No comments:

Post a Comment