Specific aim 0: Verify or deny the association of fetishism with interictal or sub-ictal epileptiform EEG in the temporal lobe.
Rationale. The association of fetishism with EEG abnormalities has a long history in the literature, but most of the information available on this topic comes from case reports and anecdotes. In fact, the largest EEG study on fetishists [Epstein 1960] isn't much more than a collection of cases similarly discussed and evaluated. In the converse domain of fetishism studies on epileptics there has been only one entry [Kolársky & al. 1967]. Although this used a clinical population of a respectable size (86), paring it down according to location of the epileptic focus left only 49 temporal-lobe cases. Furthermore, this study addressed `sexual deviation', a heterogeneous and culturally defined phenomenon of which fetishism is only one particular type, and the diagnostic criteria were not rigourously specified. The `deviant' group included not only fetishists but also voyeurs, exhibitionists, sadists, masochists, pedophiles, and, most inappropriately, homosexuals. Most of the literature on sexual deviation suffers from this problem of inclusivity. In any study of a heterogeneous phenomenon of which the biological basis is uncertain, it is best to opt for specificity in diagnostic criteria, as long as an adequate sample size can be obtained. This prevents the masking of diverse biological substrates. The conclusions of these previous studies must be regarded with suspicion until they are verified in studies with narrow inclusionary criteria and large sample sizes.
Given this desirability of narrow criteria, it may seem counterproductive to lump the distinct DSM-III-R diagnoses of fetishism and transvestic fetishism into a single study group. On the contrary, these diagnoses are so similar in terms of motivation, diverging only in the specific nature and application of the fetish object, that merging them in order to increase the size of the study population is acceptable.
Methods. In order to establish the generality of any conclusions, on the order of a hundred subjects will be needed. These will be gathered from responses to advertisements posted in fetish-oriented clubs, shops, and publications, and also on local Internet sites. Any subjects taking psychotropic medication will be excluded. Informed consent will be obtained. Subjects will be assessed for DSM-III-R axis I disorders using the SCID-NP with the abbreviated B/C module for psychotic screening. Subjects will be assessed for DSM-III-R axis II disorders using the SCID-II. All subjects having a current or historic non-paraphilic axis I disorder (including obsessive-compulsive disorder, which the DSM-III-R distinguishes from the paraphilic disorders) will be excluded. All subjects having a current or historic axis II disorder other than obsessive-compulsive personality disorder will be excluded. (As noted above, reports suggest that obsessive-compulsive personality traits are common in fetishists.) Fetishism (DSM-III-R 302.81) or transvestic fetishism (DSM-III-R 302.30) will be confirmed by diagnostic interview, and the features of the fetishistic sexual outlet will be recorded. These features include the nature of the fetish object or objects, routines associated with the fetish, any obsessive or forced appearance of the fetish in consciousness, amount of time taken up by fetishistic behaviour, distress caused by the fetishistic behaviour, and prevention or interference with reciprocal affectionate sexual activity. To minimise confounds, fetishists with a Hamilton Depression Rating Scale [Hamilton 1960] score greater than 7 will be excluded. History will be taken with an eye toward identification of any possible brain trauma. (No assumptions are made about the expectation of finding such trauma, but the possible relationship of fetishism to abnormal EEG is reason enough to keep an eye open for it.) Birth history will be particularly useful in this evaluation, but cannot be an absolute requirement of the study since it is often not readily available in adult populations. An age-matched control group will be selected from the respondents whose diagnostic interviews rule out fetishism and transvestic fetishism, and who evidence no other psychiatric or neurological abnormality. Selecting the controls in this manner, from the same group of respondents, will help to limit the divergence between groups to the diagnosis of fetishism.
Subjects will be instructed to think of their fetish object or objects during the recording session. Controls will be instructed to think of a favourite sexual prop or partner. EEG with 15s[-1] photic stimulation and an initial three-minute period of hyperventilation, which are effective activation procedures for the detection of spike activity [Striano & al. 1992], will be recorded for a total of fifteen minutes in a temporal-lobe montage consisting of the eight bipolar channels F7-T1, T1-T3, T3-T5, T5-P3 (left), and F8-T2, T2-T4, T4-T6, T6-P4 (right) of the international 10-20 system. (T1 and T2 are interpolated halfway between F7 and T3 and F8 and T4, respectively, which places them near the temporal pole, almost overlying the Sylvian fissure. Sites T1 and T2 are as sensitive as nasopharyngeal electrodes to temporal-lobe spiking [Dave Woods, personal communication], and are of course much more comfortable.) A bipolar montage makes sense when one is interested in detecting spikes originating from definite foci, because polarity inversions associated with the traversal of the focus can be detected in the channels that share a pole near the focus. Such an arrangement also minimises contamination from photomyoclonus.
It is realistic to predict a processing rate of five EEGs per week. Diagnostic assessments can be pipelined with EEG, so that the whole process of diagnosis and EEG recording for a group of a hundred subjects and an equally sized control group could be accomplished in one year.
Analysis. Computer algorithms for automated spike detection are beginning to become available [Gabor & Seyal 1992, Jandó & al. 1993], but they suffer from an unacceptably high rate of false identifications. Until such algorithms have been improved and validated, experiments will have to rely on human judges. Accordingly, the EEGs will be scanned for spike events by at least two judges independently. Judges will be blind to the group identity of each EEG record. In order for a spike to be counted, all judges must identify it. In addition to the number of spike events, the site and laterality of any focus will be noted.
The analysis of the study group in comparison to the control group is critical to the interpretation of any result. This may seem so obvious that it isn't worth stating, but in the past the absence of controls has compromised the validity of EEG studies on sexual deviants. EEG spiking does occur in behaviourally normal populations with an incidence probably between 0.3% and 3% [Chatrian 1976, Binnie 1988], so what is important in this analysis will be the difference in spike incidence between fetishists and normals. For each of these groups, two measures will be computed: the mean number of spikes per subject, and the number of subjects manifesting spikes. The difference in these measures between groups will be tested for significance.
Interpretation. A failure to replicate the historical finding of an increased incidence of temporal-lobe EEG spikes in fetishists would suggest that the impression of such a relationship was the result of a reporting bias; a case of fetishism with EEG abnormality is more noteworthy than one without. However, a negative finding on fetishists would of course leave open the possibility that EEG abnormalities might be associated with some of the other abnormalities of sexual outlet that have been lumped together as `deviance' in previous studies. Also, the association could be present, but exaggerated by reporting bias so that the sample of the current study was too small to catch it. Finally, the a single session of surface recording might be too insensitive to detect an abnormality.
Replication of the finding would suggest an exciting correspondence between deviation in neural activity and difference in behaviour, although it should not be too eagerly interpreted. A correlation of two phenomena offers two main possibilities: one, that one of the phenomena causes or contributes to the other, and two, that both phenomena are products of some antecedent causal factor.
The most interesting interpretation would of course be that EEG spikes are causally related to fetishism at the neural level. The abnormal synchronous firing of large groups of cells might interfere with the formation, maintenance, or application of the associations necessary for a normal sexual releasing mechanism, or might induce new associations that supersede the normal ones. The effect of large population spikes on an intricate and small-scale neural network might be analogous to manipulating the values of large clusters of units in an artificial neural network: in an attempt to degrade gracefully, the network might produce a meaningful output, but that output might have little relevance to the situation presented in the input.
An association of fetishism and temporal EEG spikes alternatively could indicate that the same sort of trauma often produces EEG spikes and a predisposition toward fetishism. That is, the behavioural and electroencephalographic abnormalities could be products of a common cause without the EEG's having any effect on the behaviour.
Related to this last point is the possibility that EEG spikes are not directly causally associated with fetishism but are associated with other abnormal or fringe behaviours or traits, which make the individual less desirable as a sexual partner and therefore frustrate the achievement of normal sexual outlet. In such a situation, fetishism might develop as a substitute outlet, an adaptive response akin to dominance failure mechanisms [Wilson 1987] in animals. Mitigating against this interpretation are the reports of fetishism in men who are married or otherwise engaged in stable sexual relationships. These people might not always have had such outlets available, though, and fetishism once developed may be difficult or impossible to break.
Specific aim 1: Assess the effect of the serotonin selective reuptake inhibitor fluoxetine on fetishistic thoughts and behaviours.
Rationale. Within the past three years reports of successful treatment of fetishism and other paraphilias with serotonin selective reuptake inhibitors (SSRIs) have proliferated. These results are exciting, but as is the case for the reports of EEG abnormalities and fetishism, showmanship and story-telling don't count for scientific proof. Diagnoses, drugs, dosages, and analyses all have differed, populations have been small and diagnostically heterogeneous, and controls have been nonexistent. In addition, some case reports may have confounded alleviation of fetishism with a reduction in total sexual outlet, since serotonergic manipulations are known to affect normal sexual behaviour in animals [Everitt 1983]. Based on use of a new self-reporting scale that treats both paraphilic and normal sexual behaviours, the Sexual Outlet Inventory (SOI), Kafka and Prentky recently [1992] reported a dissociation between paraphilic motivation and global sex drive.
Researchers into fetishism often have remarked on its obsessive-compulsive character. At least for some subpopulation, fetishism may be no more than a particular type of obsessive-compulsive outlet. However, the usual obsessive-compulsive scales and diagnostic tools emphasise routines and belief systems such as cleaning, checking, and contamination, and this emphasis biases against identification of sexual compulsions as part of the OCD spectrum, even though they may be usefully viewed as obsessive-compulsive in nature. (This is also true of disorders other than paraphilias, for example Body Dysmorphic Disorder in which patients are obsessed with the possibility of abnormalities in physical form.)
Possession and manipulation of fetishes and performance of associated rituals are widely reported to relieve anxiety in fetishists. This suggests that fetishism may become or even develop as a mechanism for anxiety relief.
A study is needed that assesses the therapeutic efficacy of a single SSRI in a population composed exclusively of fetishists, with evaluation not only of the frequency and intensity of fetishistic sexual outlets, but also of the fetishistic outlets as a fraction of the total sexual outlet. Furthermore, this study should explicitly evaluate obsessive-compulsive mechanisms and anxiety as possible contributors to any change in sexual outlet.
This raises the question of which SSRI to use in such a study. Fluoxetine is suggested in this proposal because it is relatively well tolerated, has gained wide clinical usage, and has been the most commonly applied SSRI in case reports of paraphilias.
Methods. Subjects who are on other drugs that might interact with fluoxetine will be excluded from this phase of the study. Such drugs include all that have a high affinity for plasma protein, since fluoxetine is 94.5% protein-bound and competition for binding may displace it and thus increase its effective concentration. Any subjects with hepatic or renal disease or diabetes, or who are medicated with a drug metabolised by the cytochrome P450IID6 system which may compete with fluoxetine, will be excluded. Interaction of fluoxetine and monoamine oxidase inhibitors also is a clinical danger, but since all prospective subjects taking psychotropic medication will have been excluded during screening for the previous phase of the experiment, no such interaction will be possible.
The manufacturer recommends for depression an initial daily dose of 20mg fluoxetine with upward titration if indicated. Though 20mg, 60mg, and even 80mg doses have occurred in case reports of fetishism and other paraphilias, 40mg has been the most common dose and seems a reasonable level for the purpose of verifying the phenomenon suggested by these reports. Additionally, when Dailey & al. [1992] administered to rats amounts of intraperitoneal fluoxetine that protected against seizures, they found that the plasma levels of fluoxetine and norfluoxetine reported by the manufacturer after 30 days of daily oral dosing at 40mg in humans were 93% and 15% of the average level in the their rats.
One would hope to be left with on the order of 20 subjects from the initial study who want to reduce their fetishistic behaviours, do not meet exclusionary criteria for the drug study, and will give informed consent for the drug study. If the sample size is not adequate, more subjects will be recruited. Judging from the effect on the Sexual Interest Ratio measure of the SOI in [Kafka & Prentky 1992], which was reduced from 87.5 to 47.6 with a standard deviation of 33.1, even a final sample size on the order of 10 would be sufficient to achieve statistical power (critical z-score for p=0.05 (two-tailed) is 1.96, so (87.5-47.6)/(33.1/[[radical]]N) >= 1.96. Solving for N, we find N > 2.6 for this level of power.) Subjects will receive either 40mg fluoxetine or placebo daily for periods of 12 weeks in a double-blind crossover design, with fluoxetine and placebo separated by a washout period of 8 weeks. (A long washout period is indicated due to the long half-lives of fluoxetine and its active metabolite norfluoxetine.) Due to the possibility of adverse reactions or side effects, the first week of the treatment period will be used for upward titration of the dose. During the fluoxetine and placebo periods, subjects will complete the SOI and the A-State subscale of the State-Trait Anxiety Inventory (STAI) [Spielberger & al.], both self-report questionnaires, weekly. Every four weeks during these periods they will be interviewed for the YBOCS. The subjects's dates of starting the experiment will be staggered in order to maintain throughput. At a rate of 5 YBOCS administrations and EEGs (see specific aim 2) each week, a group of 20 subjects could be processed in 35 weeks.
Analysis. For the SOI, YBOCS, and STAI/A-State, pooled data from all subjects will be plotted for periods of fluoxetine and placebo and analysed for significant differences between the two conditions. In addition, separate analyses will be done on the subgroups who did and did not manifest EEG spikes during the initial phase of the study, and on subgroups with and without a diagnosis of obsessive-compulsive personality disorder. (Note that obsessive-compulsive personality disorder, a DSM-III-R axis II classification, is distinct from obsessive-compulsive disorder, an axis I classification in which specific, well-defined, ego-dystonic obsessions and compulsions are present and which is part of the exclusionary criteria for this study.)
Though fluoxetine is relatively well tolerated, some side effects are possible and these may prompt withdrawal from the study. Depending on the number of withdrawals, either these subjects will be eliminated from the study or their data from the latest evaluation under each treatment condition will be carried forward, ensuring that any error is in the conservative direction. Possible non-sexual side effects of fluoxetine include anxiety, insomnia, weight loss, gastrointestinal upset, and sedation.
Possible sexual side effects of fluoxetine include anorgasmia, impotence, and decreased libido. These physical and mental obstacles decrease the total sexual outlet. For this reason it is especially important to use a measure such as the SOI which expresses paraphilic sexual outlet as a fraction of the total sexual outlet.
Interpretation. A coupling of the YBOCS score with the SOI's measure of abnormal sexual outlet would be expected, since the YBOCS depends upon a priori definition of symptoms and in this study such a definition is likely to include fetishistic thoughts and behaviours as a major feature. The strength of the correlation between these two would be an index of the value of the obsessive-compulsive model as an interpretive framework for fetishism.
The value of the anxiety model can be assessed using STAI scores. If the STAI score moves in the same direction as the SOI unconventional sexual outlet and the YBOCS, then fetishism might be producing anxiety, since decreasing one decreases the other. An alternative interpretation of a positive correlation between STAI and SOI and YBOCS might be that fluoxetine's effect on fetishism is secondary to an anxiolytic effect. Such an anxiolytic hypothesis could be tested by comparing the effects on anxiety of fluoxetine and an anxiolytic such as alprazolam; indeed, Fedoroff [1988] has done exactly this in a single case (with buspirone instead of fluoxetine) and reported that alprazolam dissociates anxiolytic effect from anti-paraphilic effect.
If the STAI score moves in the opposite direction from the SOI abnormal sexual outlet and from the YBOCS sub-scores on time spent on compulsions and time spent on obsessions, then fetishism may be serving as an anxiety release. In this case, fluoxetine would be preventing the release of fetishistic behaviour without affecting the underlying anxiety or enabling a more normal means of release.
Significant differences in the data from subjects with and without EEG spikes may indicate a heterogeneity in types of fetishism, determined by their neural substrates. It is possible, for example, that fetishism is more an obsessive-compulsive behaviour in people who manifest EEG spikes and more an anxiety response in people who do not. Although the size of the spike group may not be adequate to assert the significance of any effect, it will at least be an indication for further work.
A lack of effectiveness of fluoxetine in the absence of a diagnosis of obsessive-compulsive personality disorder would support the view that the claims of anti-paraphilic efficacy of fluoxetine are an artefact generated by comorbidity of these two conditions. Again, the sample size here may or may not be adequate, depending on the extent of comorbidity, but this observation will at least be an indication for further work.
Specific aim 2: Assess the effect of fluoxetine on the EEG of fetishists.
Rationale. To put it simply, EEG spiking has been reported in fetishists, and SSRIs have been reported to relieve fetishism, and this is reason enough to examine EEG effects in any search for the therapeutic mechanism of these drugs. This phase of the study is of course contingent on the confirmation of the earlier EEG results (specific aim 0).
Methods. This part of the study will occur concomitantly with the procedures described above under specific aim 1. For any and all subjects who manifested EEG spikes during the initial phase of the study, EEG will be recorded as described for specific aim 0, every four weeks at the time of administration of the YBOCS as described for specific aim 1.
Analysis. EEG will be again be analysed as described in specific aim 0. Behavioural scores from these subjects will be analysed for covariance with number of EEG spikes.
Interpretation. Because the dynamics of seizure initiation are not understood, the relationship of ictal and interictal activity is uncertain. The various modulatory effects of serotonin in different brain regions further complicate the picture. In a very detailed study on potassium-induced spiking in rat hippocampal slices, Jensen and Yaari [1988] used cadmium to block calcium currents associated with synaptic transmission and, in a different experiment, a glutamate blocker. Both suppressed synaptic transmission and spiking. However, generation of seizures was actually accelerated by this manipulation. In another experiment, a 1s[-1] simulated spike input to the isolated CA1 field decreased seizure frequency, perhaps because of inhibitory responses to spikes. Gotman [1991] and Gigli and Gotman [1991] studied the relationship between interictal spiking and seizures in humans and found no causal correlation of spike rate with seizure frequency, though spike rate often did increase after seizures. Likewise, the reduction of many anticonvulsant drugs has no definite effect on spike rate, though it does precipitate seizures, which in turn increase the spike rate [Gotman & Koffler 1989, Sundaram & al. 1990, Duncan 1987, Duncan & al. 1989]. Furthermore, Engel and Ackermann [1980] found an inverse relationship between spike rate and seizure frequency in amygdaloid-kindled rats, and Swartzwelder & al. [1987] found likewise in magnesium-depleted brain slice model of epilepsy. The phenomenon of interictal or subictal spiking seems to some degree independent of actual seizure activity, and it could very well subtly exert influence on behaviour.
If spikes are present in the EEG of fetishists, their diminishing concomitantly with fetishistic thoughts and behaviours during fluoxetine treatment would support, but not establish, a causal relationship between the spikes and the fetishism. Even if an association is demonstrated, it still could be merely correlative and not causal. If spikes are present but do not diminish concomitantly with fetishistic thoughts and behaviours, this would establish some degree of independence of the two phenomena.
Differences between the baseline and four-week EEG recordings would be particularly important in suggesting an answer to the question of causation in the relationship of EEG spikes to fetishistic behaviours. It is well known that the time courses of the therapeutic effect and the pharmacological effect of SSRIs do not correspond. As therapies both for depression and for obsessive-compulsive disorder, SSRIs are effective only after several weeks of dosing. Even though they begin to inhibit reuptake of serotonin very shortly after administration, the full behavioural effect does not manifest until sometime around the four-week mark. If the course of change in spike rate parallels the courses of SOI unconventional sexual outlet and YBOCS scores, this would support a causal relationship between the spikes and the behaviours. If, on the other hand, the spike rate decreases more sharply than the behavioural scores at the first evaluation, this would suggest a more direct link between spiking and serotonergic uptake blockade, and a merely correlative link between spiking and fetishistic behaviours. The reliability of spike detection in any single session may, however, not be adequate for such an analysis.