CENTER FOR DRUG EVALUATION AND
`
`RESEARCH
`
`APPLICA TION NUMBER:
`
`18-936/SE5-064
`
`PHARMACOLOGY REVIEW
`
`
`
`

`

`kUMI’LCVBO "" W'V
`
`
`
`FLUOXETINE DEVELOPMENTAL TOXICITY:
`
`ANIMAL-TO-HUMAN COMPARISONS
`
`A Report
`
`Sonia Tabacova
`
`National Center for Toxicological Research / FDA
`
`General Information
`
`Drug name(s):
`
`Fluoxetine (Adofen, Fluctin, Fluoxeren, LY-110140,
`Fontex, Foxetin,
`,.,__~_
`\
`
`Structure:
`
`(:)N—methyl-3-phenyI-3-(4-trifluoromethyI-3-
`phenylmethylphenoxy)propy|amine
`
`Year approved:
`
`1988
`
`Therapeutic uses:
`
`Antidepressant (a selective serotonin reuptake inhibitor)
`The most frequently prescribed antidepressant in the U. 8.
`
`Duration of dosing:
`
`Prolonged (often throughout gestation)
`
`Pregnancy category Pregnancy Category B
`and warnings:
`Most common side effects: nervousness, anxiety, nausea,
`insomnia, anorexia, diarrhea, headache
`
`
`
`

`

`
`
`Contents
`
`FLUOXETINE DEVELOPMENTAL TOXICITY
`
`I. EVALUATION OF HUMAN DATA
`|.1. Materials
`I.2. Methods
`l.3. Results
`1.3.1. Evaluation of studies
`
`1.3.2. Adverse outcomes in relation to gestational exposure
`to fluoxetine in humans
`
`II. EVALUATION OF ANIMAL DATA
`”.1. Materials
`".2. Methods
`".3. Results
`
`ll.3.1. Effect on fertility
`ll.3.2. Prenatal develomental effects
`".32. Postnatal develomental effects
`"3.4. Maternal effects
`
`Ill. ANIMAL-to- HUMAN COMPARISONS
`
`lll.1. Pharmacokinetics and pharmacodynamics
`III.2. Reproductive and Developmental Toxicity
`
`References
`
`Annex 1: Format
`
`Annex 2: Extended summaries and evaluation of animal studies
`
`(Tables 4.1 —4.6.)
`
`000903
`
`"PP?
`
`2°
`
`2°
`
`pppppppp
`
`'P'P'P
`
`.15
`
`.17
`
`.21
`
`
`
`

`

`FLUOXETINE DEVELOPMENTAL TOXICITY: I. EVALUATION OF HUMAN DATA
`
`l.1. Materials
`
`Reviewed 'are 12 publications of studies on the outcomes of gestational exposures to
`fluoxetine
`in human subjects. Eight of these publications are population studies of
`prospective design (incl. 3 controlled epidemiological analytical studies and 5 descriptive
`uncontrolled case surveys based on information from premarketing clinical
`trials or
`postmarketing reports); three papers a re p ublications of single case reports; and one
`paper is a meta-analysis performed on all available published and unpublished reports
`(up to November 1996) on congenital malformations observed after fluoxetine use during
`the 1st trimester of pregnancy.
`
`l.2. Method
`
`The review process was performed in the following consequtive steps:
`
`1. Selecting papers for review by means of excluding irrelevant papers and those not
`containing new information, such as repetitive publications or literature reviews.
`
`2. Abstracting data from the selected publications in an uniform way, according to the
`endpoints that have been recommended to be included in a reproductive/developmental
`toxicity database (as described by Carole Kimmel in Apendix 1 of the project proposal).
`The endpoints include: type of data, number of subjects, exposure parameters (dose,
`time, and duration of exposure), other potential risk factors or confounders, effects on
`offspring (embryo/fetal or perinatal death, prematurity, congenital anomalies, altered
`birth weight, postnatal complications, developmental delays, neurobehavioral or other
`organ systems' deviations) and maternal effects.
`
`3. Evaluation of data (each study separately) according to the following criteria:
`- Reliability of study design
`- Appropriate control group
`- Sufficient number of subjects
`- Adequate assessment of exposure and outcome(s)
`- Control of potential confounding factors (eg. maternal age, gravidity, parity,
`previous adverse outcomes, pre-existing and/or concurrent disease, socio-
`economic factors, medication or drug use, smoking, alcohol consumption,
`occupation, etc).
`— Relevant statistical analysis to assess the relation between exposure and
`outcome.
`
`Additional criteria, such as: plausibility of results (having in mind the known
`pharmacokinetic and pharmacodynamic properties of the agent), and concordance with
`other studies, were introduced in the evaluation of single case reports where some of the
`above listed criteria could not be applied.
`
`4. Conclusion about data reliability for each study separately (based on steps 2 and 3).
`
`

`

`5. Overall assessment of the observed adverse effects and the likelihood of their causal
`
`relation to fluoxetine exposure during gestation, based on Hill’s criteria for establishing
`causation as follows:
`
`- Strength of evidence
`- Consistency of evidence
`- Specificity of effect
`— Temporality of effect
`— Dose-response relationship
`— Plausibility of effect
`- Coherence with existing knowledge
`- Analogy (structure activity)
`(Hill AB. The environment and disease: association or causation?
`Proc.R. Soc. Med. 1965, 58:295-300
`On the basis of this assessment, a selection of outcomes associated with fluoxetine
`
`gestational exposure in humans was made, in order to be compared with experimental
`animal data.
`
`L3. Results
`
`|.3.1. Evaluation of studies
`
`The reviews of human studies,
`
`incl. abstracted data, their evaluation, and conclusion
`
`about data reliability, are presented in Tables 1 and 1-a (for population studies and
`single case reports respectively). A short summary of the above information is presented
`in Table 2.
`
`The study design is evaluated as reliable only for the 3 epidemiological analytical studies
`on fluoxetine gestational exposure in association with signs of developmental toxicity
`(Pastuszak et al, 1993, Chambers et al, 1996, Nulman et al, 1997). These studies are of
`a prospective cohort type (the best type of design to study a cause-effect relationship),
`with adequate control groups.
`In contrast, the reliability of study design to assess a
`relationship between fluoxetine
`gestational exposure and adverse pregnancy/birth
`outcome is poor for single case reports, and limited for the descriptive case surveys
`since they lack control groups and are prone to bias because they are based on
`voluntary spontaneous case reports instead on a random sample.
`
`The sample size varies from 17 to 544 in the descriptive case surveys, and from 219 to
`482 (incl. control groups) in the epidemiological studies. For most of the descriptive
`surveys the sample size is insufficient to provide reliable estimates of the rates of
`adverse outcomes. Although the overall sample size is more adequate in the analytical
`epidemiological studies,
`it
`is insufficient
`in some stratification subgroups (eg.
`the
`subgroups of patients treated during entire gestation or 1st trimester only, Nulman et al,
`1997), or of insufficient power to detect a small or moderate increase in the risk of
`adverse outcomes (e.g. Pastuszak et al, 1993).
`
`The doses of maternal exposure to fluoxetine are not reported in 5 of the reviewed
`publications (4 surveys and 1 epidemiological study); presumably the doses were within
`the therapeutic dose range (10 to 80 mg/day), since all these studies involved subjects
`that had been treated with fluoxetine for therapeutic reasons. Where indicated, the mean
`daily doses of fluoxetine were 25 or 28 mg, and the range - from 10 to 80 mg/day.
`
`
`
`

`

`The timing and duration of exposure to fluoxetine have taken place during the 1st
`trimester of pregnancy in the majority of studies reviewed (8 of 12).
`in 2 of these studies
`(the analytical epidemiological studies of Chambers et al, 1996 and Nulman et al, 1997)
`there are subgroups that were exposed during entire gestation as well, but they were
`analyzed separately and did not confound the data pertinent
`to the 1st
`trimester
`exposure. One studiy is based on exposures during 3rd trimester to delivery (the survey
`of Goldstein et al, 1995), and in two publications the timing and duration of exposure are
`not reported.
`
`including controlling for other potential risk factors
`An appropriate statistical analysis
`that could confound the association between exposure and outcome is essential
`in
`assessing a cause/effect relationship and in evaluating the reliability of a study. The
`association between exposure and outcome taking into account potential confounders
`has been analyzed only in the 3 above mentioned epidemiological analytical studies
`(Pastuszak et al, 1993; Chambers et al, 1996, and Nulman et al, 1997). The rest of the
`studies (case surveys and case reports) are descriptive, and therefore unsuited for
`analyzing relations between exposure and outcome. Although in some of these studies
`(e.g. Goldstein et al, 1995; McElhatton et al, 1996) confounding factors were registered,
`their
`role was not
`taken into account because of
`the limitations inherent
`to the
`
`descriptive study design.
`
`In summary (table 2), out ofthe 1 2 studies reviewed, the reliability of data has been
`evaluated as ‘good’ in 6 studies, including:
`the 3 epidemiological analytical studies of
`prospective cohort design which had appropriate control groups and took into account
`possible confounding factors; 2 case reports with reliable assessments of exposure and
`outcome, findings consistent with other studies, and plausible results with regard to the
`pharmacokinetics and pharmacodynamics of the agent; and 1 meta-analysis by Addis
`and Koren, 1997 (not shown in the table). Four studies are of limited reliability (all of
`them are uncontrolled descriptive case surveys based on spontaneous reports), and the
`remaining two (a case report by Venditelli et al, 1995, and a case survey by Brunei et al,
`1994) are of poor reliability due mainly to confounding by multidrug exposures and very
`small number of subjects, respectively.
`
`l.3.2. Adverse outcomes in relation to gestational exposure to fluoxetine in
`humans
`
`Table 2 summarizes the data onapregnancy and birth outcomes according to literature
`source.
`In table 3, an attempt is made to evaluate the likelihood of a causal relation
`between the reported adverse outcomes and gestational exposure to fluoxetine, taking
`into account the compliance with Hill’s criteria for establishing causation.
`
`in
`As seen in tables 2 and 3, the reported changes in pregnancy and birth outcomes
`women treated with therapeutic doses of fluoxetine during gestation are subtle. They are
`listed below in order roughly corresponding to their occurrence in association with
`fluoxetine maternal exposure.
`
`The most consistent findings are the early postnatal complications after fluoxetine
`prenatal exposures, and particularly exposures that have taken place throughout entire
`gestation until birth. Evidence for postnatal complications is present in 6 out of 7
`available publications that have studied this endpoint. Out of these, 2 are controlled
`
`
`
`

`

`
`
`epidemiological prospective cohort studies, 2 are descriptive surveys, and 2 - single
`case reports. The reported rates of postnatal complications in these studies vary
`between 3 and 13%; rates closer to the higher limit of this range, or rates significantly
`higher than the control ones are found after exposures during the last trimester or the
`entire gestation (Goldstein et al, 1995; Chambers et al, 1996),
`in comparison to 1st
`trimester-only exposures (temporality of effect). Clinically, the postnatal complications
`are expressed
`by consistent symptoms:
`jitteriness,
`irritability, hypertonia,
`tremor,
`impending seizures,
`respiratory problems, as well as subcutaneous or
`internal
`hematomas (petechiae, cephalhematoma, or periventricular bleeding). These symptoms
`are in agreement with the known side effects of fluoxetine in adults, as well as with the
`known pharmacokinetic and pharmacodynamic properties of fluoxetine,
`including its
`effect on the platelets. The disappearance of these symptoms in the neonate is
`concordant with the decrease of fluoxetine and its main metabolite in cord blood. These
`data testify to a coherence with the existing knowledge, plausibility, and specificity of
`effect. Therefore, although the strength of evidence from a statistical point of view is
`limited (due mainly to insufficient controlled epidemiological studies on this endpoint), a
`causative relation to fluoxetine exposure is strongly suggested.
`
`Data on spontaneous abortions after gestational exposures to fluoxetine are available
`in 6 studies, 3 of which are analytical epidemiological studies and 3 — descriptive
`surveys.
`in 5 of these 6 studies,
`the reported spontaneous abortions' rates are quite
`consistent (within the range of 13 to 15.9 per cent of all pregnancies exposed
`to
`therapeutic doses of fluoxetine during the ist trimester). These rates were found to be
`higher in comparison to control
`in 2 out of the 3 analytical epidemiological studies
`(Pastuszak et al, 1993; Nulman et al, 1997), but this difference did not reach statistical
`significance due mainly to the insufficient number of subjects. Similar rates are reported
`in most of the descriptive surveys (Goldstein and Marvel, 1993; Shick-Boschetto et al,
`1992; McElhatton et al, 1996), but the absence of control groups in these descriptive
`studies does not allow to assess the risk.
`
`Having in mind the methodological uncertainities inherent in determining the true rate of
`spontaneous abortions (Wilcox, 1991), the available data provide unconclusive evidence
`for either supporting or refuting an association of fluoxetine gestational exposure with
`spontaneous abortions.
`
`Prematurity rates after fluoxetine use in pregnancy are reported in 7 of the reviewed
`population studies (3 epidemiological and 4 surveys). Out of these, a statistically
`significantincrease in prematurity rate is found in only one study,
`in association with
`continued fluoxetine use after 25 week gestation ( Chambers et al, 1996). This is a
`controlled prospective cohort epidemiological study of good reliability. However,
`this
`finding is not confirmed by any other studiy. This lack of consistency could be due to the
`fact that in the majority of the other studies the timing of exposure was different: it took
`place earlier (during the tst trimester)
`as compared to 3rd trimester and later in
`Chambers’ study. In support,
`in that same study, maternal exposures prior to 25 weeks
`of gestation did not result
`in increased prematurity rates.
`it
`is also possible that
`Chambers’
`findings could have been confounded by parallell exposures to other
`psychoactive drugs that took place in 30 per cent of the patients.
`
`The evidence is insufficient to draw a definite conclusion.
`
`

`

`Data on birth weight are available in 4 population studies (including 3 epidemiological
`and 1 case survey) and in 3 case reports. These studies provide consistent evidence for
`a lack of effect on birthweight after fluoxetine maternal exposure during the first trimester
`of pregnancy.
`Altered birthweight (a statistically significant decrease) is found in association with ‘late’
`exposures to fluoxetine continued after 25 weeks of gestation in an epidemiological
`study of good reliability(Chambers et al, 1996). This finding is plausible and coherent
`with the existing knowledge about the anorexic (apetite suppresing) effect of fluoxetine in
`therapeutic doses in adults. Thus, the decrease in birthweight could be secondary to a
`diminished maternal
`food consumption. Unfortunately,
`the study does not provide
`information on the maternal effects. A causative relation is possible, but the conclusion
`is hindered by the lack of other studies on birthweight after ‘late' gestational exposures
`to fluoxetine.
`
`Congenital malformations. There is a consistent evidence in all studies for a lack of
`any increase in major congenital malformations, or for a pattern of malformations
`associated with exposures to fluoxetine during gestation. Two epidemiological studies
`(Pastuszak et al, 1993 and Chambers et a, 1996) report an increase in minor congenital
`malformations in comparison to control. The difference reaches statistical significance in
`only one study (Chambers et al, 1996), however no description of
`the 'minor'
`malformations found is provided. The prevailing evidence is in favor of a lack of relation
`of fluoxetine gestational exposure to birth defects. This conclusion is supported by the
`meta-analysis of Addis and Koren (1997) which takes into account all available
`information on fluoxetine 1st trimester exposures up to 1996.
`
`Perinatal death. There is consistent evidence (6 out of 6 population-level studies) that
`maternal exposure to therapeutic doses of fluoxetine is not associated with increase in
`perinatal lethality.
`
`Neurobehavioral development of children prenatally exposed to fluoxetine has been
`followed
`up
`in
`only
`one
`study
`(Mattison
`et
`al,
`1999),
`a
`comprehensive
`neuropsychological evaluation of 66 children at 4-6 years of age, born to prospectively
`identified women who were taking fluoxetine during pregnancy,
`in comparison to a
`prospectively identified control group of 30 children of mothers with pregnancy
`exposures to drugs “not deemed to be teratogenic”. Verbal
`learning and memory,
`language, short-term memory/attention, motor, parent-rated behavior, and IQ scores
`were analyzed and
`compared using appropriate statistical
`tests. No statistically
`significant group differences were found,
`suggesting that
`the risk of negative
`neurobehavioral outcome in fluoxetine—exposed children is similar to that of non-exposed
`ones. The lack of supportive evidence from other neurobehavioral studies does not allow
`a definitive conclusion about this endpoint.
`
`
`
`
`
`

`

`
`
`II. EVALUATION OF ANIMAL DATA
`
`".1. Materials
`
`Reviewed are all available on file animal reproductive and developmental toxicity studies
`on Fluoxetine safety assessment, as follows:
`two reproductive/fertility studies in the rat
`(also involving teratology and postnatal segments) and four prenatal developmental
`toxicity studies (including two preliminary and two routine teratology studies in two
`animal species - rat and rabbit). The reviewed studies are listed below:
`
`1. A Fertility Study on Fluoxetine Hydrochloride (LY110140) in the Female Rat (1980).
`Laboratories Study No R0 7179 by J. Wold , N. Owen and E. Adams
`
`Lilly Research
`
`2- A Fertility Study, including Behavioral and Reproductive Assessment of the F1 Generation, in the Wistar
`Rat Given Fluoxetine Hydrochloride (LY110140) in the Diet (1982 ). Lilly Research Laboratories Study No
`R10280 & RO4781 by G.Brophy, N. Owen and J. Hoyt
`
`3- A Preliminary Teratology Study on Fluoxetine (Lilly Compound 110140) in the Rat (1979). Lilly Research
`Laboratories Study No R-77, IND '-- Toxicology Report No 7 by J.S Wold and J.K. Markham
`
`4. A Teratology Study on Fluoxetine (Lilly Compound 110140) in the Rat (1979). Lilly Research Laboratories
`Study No R-207, IND N .Toxicology Report No. 8 by J. S. Wold and J. K. Markham
`
`5- A PreliminaryTeratology Study on Fluoxetine (Lilly Compound 110140) in the Rabbit (1979). Lilly
`Research Laboratories Study B -7017,
`lND ~— Toxicology Report No. 9 by J. S. Wold and J. K.
`Markham
`
`6- A Teratology Study on Fluoxetine (Lilly Compound 110140) in the Rabbit (1979). Lilly Research
`Laboratories Study 8-7087, lND -—~ Toxicology Report No. 10 by J. S. Wold and J. K. Markham.
`
`".2. Method
`
`to human data, the experimental studies carried out to assess
`Prior to comparing animal
`safety of the agent have been evaluated.
`
`The evaluation procedure encompassed the following consequtive steps:
`1. Data collection and review: Abstracting, summarizing, and evaluating reliability of
`data from each individual study in order to identify and assess “signals" of reproductive/
`developmental toxicity;
`2. Comparison of outcomes across studies according to category (reproductive,
`developmental, prenatal/postnatal) and subclass of toxicity (fertility, embryo/fetal loss,
`dysmorphogenesis, alterations to growth, viability and functional toxicities);
`3. Evaluation of validity and reliability of identified effects (outcomes) for comparison
`with those in humans.
`
`Step 1
`The data were abstracted in an uniform way, according to a common format in order to
`facilitate data assessment and comparison across studies. The format
`(Annex 1) was
`prepared on the basis of the endpoins outlined in the Project Proposal (Kimmel et al,
`
`

`

`1997) and in accordance with the format of the National Toxicology Program’s Special
`Reproductive Study (Chapin and Sloane, 1997).
`The format consists of the following parts:
`- Data entries: particulars of animal model, exposure (compound, dose, route
`and mode of administration, timing and duration of treament) and outcomes
`(general and reproductive toxicity endpoints, subdivided into fertility, prenatal,
`and postnatal components);
`- Data summary: highlights the most sensitive endpoints, LOAEL and NOAEL
`for general, maternal, reproductive and developmental toxicity;
`Study conclusions;
`Confounding and interfering factors that might have compromised the validity
`of study conclusions;
`Evaluation of reliability of each individual study with regard to extrapolating
`the data to humans.
`
`—
`-
`
`-
`
`The evaluation of studies with regard to their reliability for extrapolating the data to the
`human was performed according to the following criteria:
`- Adequacy of experimental model
`- Adequacy of dose and route of administration
`- Adequacy of timing and duration of exposure
`- Sufficient number of animals per group
`- Presence of a dose/effect relationship
`- Appropriate statistical analysis
`- Confounders
`
`- Concordance of findings with pharmacokinetic and pharmacodynamic
`properties of the agent
`- Consistency of findings with other experimental studies
`
`involved comparison and assessment of effects across studies according to
`Step 2
`category of outcome. For this purpose, condensed comparative summaries and a
`parallel
`layout of reviewed data
`by category of outcome (e.g.
`Fertility, Prenatal,
`Postnatal) and subclass of toxicity, were prepared on the basis of the information
`collected in step 1.
`
`Step 3 involved evaluation of outcomes (presence or absence of effect on each of the
`reviewed endpoints) and their respective NOAELs in order to estimate the strength of
`evidence for each “signal of toxicity” (or no toxicity)
`and its relevance to human
`situation. Each outcome was evaluated by criteria similar to those applied above for
`evaluation of individual studies. The criteria include:
`
`- magnitude of effect (incidence relative to controlO
`-
`consistency of effect across studies
`-
`consistency of effect across species
`-
`statistical significance of effect
`-
`dose-dependence of effect
`-
`influence of confounding and interfering factors
`-
`plausibility of effect with regard to pharmacokinetic and pharmacodynamic
`properties of compound.
`
`
`
`

`

`
`
`
`".3. Results
`
`Extended summaries of fluoxetine individual experimental studies abstracted according
`to the format, along with conclusion, comments, confounders, and evaluation ofeach
`study are presented in Tables 4.1. to 4.6.
`
`Condensed comparative summaries and evaluation of fluoxetine effects by category
`(fertility, prenatal and postnatal development) are presented in tables 5.1, 5.2, and 5.3
`respectively.
`
`The endpoints (outcomes) of fluoxetine reproductive and developmental toxicity and the
`respective NOAEL levels for each outcome are summarized and evaluated in Table 6.
`
`
`All reviewed studies were performed in vivo, in animal models adequate for assessment
`of the predictive value of animal testing for human developmental toxicity.
`
`Of the total of 6 studies, 4 were performed in the rat (of Wistar and Fischer 344 strain),
`and 2 in the rabbit (Dutch Belted).
`
`The d oses a nd route of a dministration a re a dequate to h uman e xposures (oral, d ose
`range from 1.3 to 15 mg/kg/day, i.e. from a level equal to the upper limit of the human
`therapeutic dose (approximately 1 mg/kg/day) to 15 times higher. All of the studies
`employed multiple dosing regimens which allowed assessment of dose-effect and dose—
`response relationships. The mode of treatment was predominantly by gavage, with the
`exception of one study (A Fertility Study, Including Behavioral and Reproductive Assessment of the F1
`Generation, in the Wistar Rat Given Fluoxetine Hydrochloride in the Diet . Lilly Research Laboratories Study
`No R10280 8. RO4781 by Brophy, Owen and Hoyt, 1982), which employed dosing through diet
`that might have confounded the estimate of actual dose.
`
`The timing of exposure in 4 of the 6 studies covers the period from implantation to the
`end of organogenesis;
`in the remaining 2 (fertility) studies, along with gestational
`exposures, parental exposures (maternal or maternal+paternal) prior to gestation as well
`as postnatal exposures during lactation were applied. It should be noted that exposure of
`both parents is unlikely in human situation. The timing of exposure is relevant to that in
`the human studies.
`
`The number of animals/litters tested per dose group is sufficient, with the exception of
`the two preliminary, dose-finding studies in rat and rabbit (see Materials, studies No 3
`and 5).
`In the studies involving postnatal assessments (Materials, studies No 1 and 2),
`the litters were culled to a specified number only in study 2. This, along with the
`differences
`in exposure (maternal versus both parents) and mode of
`treatment
`interferes with comparability of results between these two studies.
`
`Details on the confounding and interfering factors by study and outcome are given in
`tables 4.1 to 4.6, and in table 6 respectively.
`
`10
`
`
`
`

`

`//.3. 1. Effect on fertility
`
`Fluoxetine effect on fertility (Tables 4.1, 4.2, and 5.1) was assessed in two studies (an
`one— and a two-generation study), both performed in the rat (Wistar). Both these studies
`employed similardose ranges and routes of exposure, but differred by the mode of
`treatment (by gavage vs through diet) and by the type of parental exposure (maternal
`only vs both parents). Maternal exposures took place 2 or 3 weeks before mating and
`throughout breeding, gestation and lactation; paternal exposure started at adolescence
`and continued
`for 10 weeks prior
`to mating. Both studies resulted in
`similar
`conclusions:
`fluoxetine induces no significant effect on fertility even at doses that
`produce significant general effects (NOAEL for fertility 7.4 to 12.5 mg/kg/day, as
`compared to NOAEL of 3.1 to 5 mg/kg/day forgeneral effects). The conclusions are
`reliable, although exposure quantitation may not have been precise in one of these
`studies due to dosing through diet. The general effects (decreased food consumption
`and body weight) are not necessarily a sign of toxicity, as they are characteristic of the
`pharmacological action of this drug. Statistically non-significant, but dose-dependent,
`signs of effect on fertility (decreases in fertility index, in the number of corpora lutea, in
`litter size, and increase in pre-implantation embryolethality) are found at NOAEL doses
`of 7.4 to 12.5 mg/kg/day.
`
`ll. 3. 2. Prenatal develomental effects
`
`The prenatal effects of fluoxetine exposure in utero are assessed in all 6 studies (Tables
`4.1—4.6, Table 5.2, and Table 6). Prenatal developmental effects are observed at dose
`levels that induce maternal effects (weight loss and decreased food consumption), and
`involve mostly
`an increased incidence of
`resorbed or aborted conceptuses
`(postimplantation losses) at doses of 12.5 to 15 mg/kg per day applied during
`organogenesis in two animal species (rat and rabbit). Although not consistent in the rat,
`this effect is consistent and better expressed in the rabbit. The validity of this adverse
`outcome is supported by the decrease in litter size, which, although not statistically
`significant,
`is dose-dependent and consistent across all reviewed studies in the two
`species. Fetal weight
`is usually unchanged (except for one study in the rabbit, where a
`statistically non-significant, but dose—dependent reduction by about 10% was found at 15
`mg/kg per day).
`It should however be noted that
`the effect of
`fetal weight
`is
`underestimated due to the smaller litter sizes at higher exposures in all studies (Table 6).
`Increase in congenital malformations rates is not found in either species, even at doses
`that cause maternal mortality (15 mg/kg/day in the rabbit and 40 mg/kg/day in the rat ).
`An elevation in the incidence of skeletal variations (rudimentary and wavy ribs) was
`found in one study (rabbit) at all exposure doses (2.5 to 15 mg/kg per day during
`organogenesis), but the effect was not dose—dependent.
`The NOAEL for prenatal developmental toxicity is 5 to 7.5 mg/kg per day in rat and
`rabbit respectively.
`
`ll. 3. 3. Postnatal developmental effects
`
`The postnatal effects of fluoxetine exposure are assessed in 2 studies performed in one
`animal species (rat, Wistar) (Tables 4.1, 4.2). Both studies involved oral parental
`exposures prior to pregnancy, as well as throughout entire gestation and lactation.
`However,
`the studies differred by type of exposure (maternal only vs maternal
`
`11
`
`
`
`

`

`Adverse postnatal
`+paternal) and mode of treatment (by gavage vs through diet).
`effects are found in both of these studies but are more manifested in the one that
`
`involved treatment of both parents through diet. (Tables 5.3. and 6). These effects
`involve elevated perinatal mortality (increased incidence of stillbirths and decreased
`postnatal survival during 15’ week of life), decreased birthweight and postnatal weight
`depression detectable until maturity. With the exception of stillbirths, these findings are
`reported consistently in both studies, although the changes are statistically significant in
`the above mentioned study only. The NOAEL for postnatal manifestations of
`developmental toxicity in the rat
`is from 3.1 to 5 mg/ kg per day, which suggests that
`adverse postnatal effects are induced by lower exposures in comparison to those
`inducing prenatal manifestations of developmental toxicity.
`It should be noted however,
`that NOAEL at 3.1 mg/kg is very likely to be an underestimate of the actual dose. and
`the apparent “selective" effect of
`fluoxetine on postnatal development at doses
`seemingly
`lower than those affecting prenatal endpoins may actually be due to
`exposure misquantitation because of the dosing through diet and the two-fold increase
`of maternal food consumption during lactation, as reported in that particular study.
`
`Behavioral testing of the progeny is performed in one of the two postnatal studies.
`Tested were some sensory and motor coordination functions (auditory startle reflex,
`visual placing response, rotating rod performance, and poke-hole test). No motor and
`sensory-motor behavioral deviations were found, but the testing was performed close to
`maturity (at the age of 2 to 3 months) so that earlier behavioral deviations might have
`been omitted. No detectable effect on reproductive function of the progeny was found
`
`ll. 3. 4. Maternal effects
`
`The effect of fluoxetine on the maternal organism is determined in 5 studies (tables 5.2.
`and 6). The most common effect,
`found in all studies,
`is
`the decrease in food
`consumption at dose levels of 5 and more mg/kg per day in the rat, and at lower doses
`(down to 2.5 mg/kg/day) in the rabbit.
`It is accompanied by a maternal weight loss and
`reduced gestational weight gain of about 10% during treatment. As the anorexic
`(appetite-suppressing)
`effect
`and
`the
`resulting weight
`loss
`are well
`known
`pharmacodynamic features of fluoxetine in both human and animal species,
`the
`maternal weight loss can not be interpreted as a sign of maternal toxicity, unless it is
`accompanied by other clinical signs of toxicity. Such signs are not reported, except at
`much higher doses (15 mg/kg/day in the rabbit and 20 mg/kg/day in the rat) which cause
`excessive (up to 90%) depression in food consumption and substantial weight loss
`accompanied by elevated maternal mortality.
`
`The NOEL level for maternal effects in the rat vary from 3 to 5 mg/kg/day in the different
`studies, and in the rabbit it
`is below 2.5 mg/kg/day. This shows that effects in the
`progeny occur at dose levels that cause maternal (although not necessarily adverse)
`effects.
`
`Summary
`
`It is evident that prenatal fluoxetine exposure induces pre- or postnatal developmental
`toxicity only at levels that affect the maternal organism, which means that fluoxetine is
`not a selective embryo-fetal toxicant.
`
`12
`
`

`

`toxicity in animal
`The manifestations of fluoxetine reproductive and developmental
`models, along with their respective no-effect levels are summarized in Table 6. The
`endpoints within each of the effect categories are evaluated according to their rate of
`occurrence, consistency across studies and species, statistical significance, dose
`dependence, interference of confounding factors, plausibility and coherence with existing
`information. The most “reliable” endpoints are highlighted.
`
`The outcomes for comparison with human studies are briefly outlined below:
`
`Toxicity
`
`“Positive”
`(Presence of effect)
`
`
`
`consumotion
`
`Probable
`(Effect probable)
`
`“Negative”
`(No effect)
`
`
`
`and weiht cain
`
`Congenit. Malformations
`Emb o—fetal loss
`Decreased litter size
`—— S