Sleep Medicine Reviews 18 (2014) 191e193
Contents lists available at ScienceDirect
Sleep Medicine Reviews journal homepage: www.elsevier.com/locate/smrv
GUEST EDITORIAL
Insomnia research: 3Ps and beyond Introduction The four studies featured in this volume stand on the forefront of a new generation of insomnia research. Three of the four articles [1e3] bridge the past, present, and future of insomnia research by invoking the 3P model. This model, otherwise known as the Spielman Model [4], is a time-honored framework that has been exceptionally useful for the conceptualization of factors that contribute to the development and maintenance of insomnia disorder. Both of the genetics papers [1,3] focus on the predisposition for insomnia (both in terms of behavioral and molecular genetics). One of the two genetics papers [1] raises the interesting possibility that the perpetuation of insomnia may be ascribable to not only behavioral factors (e.g., sleep extension and the mismatch between sleep ability and sleep opportunity), but also to the epigenetic alteration of biologic processes. In addition to these foci, one paper [2] offers a description of an alternative approach to the assessment of hyperarousal and the measurement of cortisol. The other paper [5] offers a potential resolution to a long-standing issue: “Do patients with insomnia, as compared to good sleepers, exhibit sleep architectural abnormalities?”. The strengths of each of the four articles are highlighted below. This is followed by the suggestion that “before moving forward, it may be best to take a step back”. That is, our research efforts may be best served by (as the geneticists would say) making sure that the phenotype of interest is well characterized. Such an effort may be the only way to properly engage with, and learn from, the next generation of methods and measures. Brief review of article findings and positions Harvey and colleagues focus on the search for, and identification of, gene anomalies that may contribute to the pathophysiology of insomnia. They propose that a genetic vulnerability is conferred by 5HTTLPR, a serotonin transporter polymorphism that regulates serotonin in the synapse. Alterations in serotonin transport (reuptake from the synapse to the presynaptic component), are hypothesized to be related to disrupted sleep via an increase in stress reactivity. They further recognize the relevance of a genetic predisposition not only for new onset of insomnia, but also in the predisposition to the conditioned effects of insomnia. For example, after a few “bad nights” it is speculated that individuals with a genetic vulnerability will be less likely to recover compared to those without such vulnerability. Apart from the specifics of what may be learned from the genetic perspective, it should also be noted that the approach is valuable because it allows for a wholly different way of thinking, in that causal inferences can be made inductively from the genetic to the 1087-0792/$ e see front matter Ó 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.smrv.2014.01.003
biological to the psychological (5HTTLPR / serotonin dysregulation / stress reactivity / insomnia). This is in stark cont rast to the hypo-deductive approach (insomnia / hypera rousal / HPA activation / hypercortisolemia) which has been typical of insomnia research in the past. Further, this approach may yield novel perspectives on which biological processes and behaviors to target for the development of new therapies. For example, in the present case, 1) the 5HTTLPR finding suggests that modulation of serotonin may be as useful a strategy as the modulation of GABA with benzodiazepines and 2) psychological interventions for insomnia may do well to adopt therapies that specifically target stress reactivity. One such intervention, for which there is already good evidence, is the addition of mindfulness training to standard CBT-I. Palagini and colleagues’ paper is an ideal complement to the Harvey paper in that it not only focuses on the genetics of insomnia, but also extends that focus to include epigenetics. With respect to the genetics of insomnia, the authors provide estimates which indicate that 1) 35e73% of patients with insomnia report familial incidence with the mother being the most frequently affected relative, and 2) twin studies show heritability ranges from 21 to 55% depending on the composition of the sample (male/female; insomnia subtype [early, middle, late]; age of onset of insomnia). With respect to the epigenetics of insomnia, the authors first provide an explanation of the phenomenon and then go on to illustrate how it may be relevant to the predisposition toward, and perpetuation of, insomnia disorder. As defined in the paper, epigenetics refers to ‘the mechanisms by which long-term programming to genes may be altered and can provide for changes in gene function without changing the gene sequence’ (p. 5). This allows for the possibility that environment can modify the genetic predisposition for insomnia, both in terms of new onset and recurrence of the disorder. Further, epigenetic mechanisms may also account for intergenerational transmission. Unlike genetics, which has a more singular focus on risk and the permanent abnormalities that are associated with chronic illness, epigenetics has a focus on the possibility that environmental factors may contribute to both the development of, and recovery from, chronic illness. Elder and colleagues argue that the cortisol awakening response (CAR) holds the promise of providing an objective measure of a biological process that may serve to predispose, precipitate, and/or perpetuate insomnia. The CAR is the hypersecretion of (spike in) cortisol that occurs with the morning awakenings that terminate the sleep period. The spike begins with the final awakening and reaches its peak about 30 min later. The authors discuss best approaches to the assessment of the CAR. The method they recommend involves sampling immediately upon awakening and every
192
Guest editorial / Sleep Medicine Reviews 18 (2014) 191e193
15 min for an hour, on two consecutive days, while controlling for potential confounds (e.g., light, saliva contamination). This empirically-based protocol could be undertaken either in the laboratory or in ambulatory settings. Finally, the authors specifically recommend the study of the CAR in relation to factors associated with the 3P model of insomnia. For example, they suggest that the CAR could provide an index of the factors that: predispose individuals to insomnia (e.g., neuroticism, etc.); precipitate insomnia (e.g., life stress or perceived stress); and perpetuate insomnia (e.g., rumination or “caffeine use”). Baglioni and colleagues provide, by way of a meta-analysis, a summary of sleep continuity and architectural differences among subjects with insomnia (PIs) and good sleepers. These data represent the most definitive work within this arena. They provide a basis on which to judge PSG findings with regard to insomnia, both with respect to individual studies and group profiles. Significant findings were detected for both sleep continuity and sleep architecture, given data from a baseline night and from night 2 studies. With regard to sleep continuity, as a group, compared to good sleepers, subjects with primary insomnia had approximately: a 6 min longer sleep latency, six more awakenings per night, 22 min more wake after sleep onset, 23 min less total sleep time, and 7% lower sleep efficiency. With regard to sleep architecture, as a group, compared to good sleepers, subjects with primary insomnia had approximately 20 min less time in slow wave and 11 min less REM sleep. The authors suggest that these differences may have implications for cognitive (memory consolidation) and emotional (stress reactivity) processing. Further, it is suggested that the PSG findings are consistent with the hyperarousal model of insomnia, at least to the extent that the significant reductions in slow wave and REM sleep represent shallow sleep. Commentary Taken together, these four articles provide clarity with respect to basic PSG findings with insomnia, highlight the prospects for the neuroendocrine study of sleep continuity disturbance, and provide an overview of the prospects for genetic and epigenetic research for insomnia. These areas of research, along with other approaches (e.g., natural history studies, longitudinal assessments of nightto-night variability, stress reactivity protocols, etc.), and other methodologies (e.g., anatomical and functional imaging, imaging of neurotransmitter “levels”, etc.) suggest that insomnia research is poised for a new beginning. Prior to engaging in the next wave of research, it seems appropriate to revisit the problems that have potentially limited our findings to date. First, as genetic research gets underway and it becomes increasingly clear that there is a need for well characterized phenotypes, it is incumbent on us to revisit the long-standing issue regarding the heterogeneity of insomnia. As alluded to by two of the current authors [3,5], certain subtypes (and types) of insomnia may reflect not only different etiologies, but also inherently different pathophysiologies. Failing to appreciate this diversity may result in a failure to detect findings. This scenario is perhaps easiest to understand when considering insomnia subtypes. A classic example where sample homogeneity may have clarified findings comes with the work of Vgontzas and colleagues [6], with respect to cortisol hyper-secretion in patients with insomnia. In this classic study, cortisol was sampled on a 30 min basis and was shown to have a circadian distribution in both good sleepers and in patients with insomnia. The groups differed across the 24 h day in only one “zone”, during the sleep onset period (approximately 22:00 he 02:00 h). The authors do not interpret this effect as being attributable to sample composition. It is possible, however, that their sample was reliably populated by patients with sleep onset insomnia, and
that some large percentage of subjects did not have middle or late insomnia. This scenario may have allowed for the detection of what was otherwise a localized finding. Unlike the given scenario, it is not hard to imagine what typically occurs with non-specific recruitment. If it had been the case that the sample was equally populated by patients with only initial, or middle, or late insomnia; and if cortisol levels were otherwise near normal or potentially supra-normal (given within-night recovery); then, in smaller sample sizes with predictably large estimates of variability, one would expect to find no difference between good sleepers and the aggregate group of subjects with insomnia. Alternatively, significant differences might be found given a large enough sample, but these differences would likely be of small magnitude and appear trivial. A similar argument may be made for the insomnia types. Presumably, the diversification of so-called “primary insomnia” into specific diagnostic entities was undertaken because it was genuinely believed that not all insomnias are the same. That is, the insomnias fundamentally differ with respect to the nature, the developmental course of, and/or the factors that contribute to the chronicity, severity, and/or frequency of, the presenting complaint. While the notion that insomnia is not a singular entity is not new (with descriptions of insomnia types and subtypes dating back to at least the 1800s [7]), the framers of the ICSD-2 apparently believed in this so strongly that they put forward no less than four separate classifications for primary/chronic insomnia (psychophysiologic, paradoxical, idiopathic, and inadequate sleep hygiene insomnia). If these different forms of insomnia have different pathophysiologies, where some of the insomnias have some features while others do not, then the likelihood of observing significant differences between good sleepers and composite samples of patients with insomnia will again, be low. For example, if hypercortisolemia is a fundamental component of psychophysiologic insomnia, but not idiopathic insomnia (which may be related to inherent neurobiologic defects), then having a sample comprised of both types of subjects is not likely to separate between good sleepers and patients with insomnia on cortisol measures, except in large samples where the finding will appear small in magnitude. As said from the outset, it seems that our best path forward may be to take a step backwards. As we are poised to benefit and learn from new methods and measures, our best strategy moving forward may be to adopt the given classifications for type and subtype and to carefully profile all research samples in these terms. At minimum, this could simply be a few rows in a descriptive statistics table indicating what proportion of the sample met criteria for the various types and subtypes. At maximum, this could entail the provision of a supplemental data table providing means and standard deviations by group and subgroup for the dependent variables of interest. With time, such systematic tracking of subject characteristics and sample compositions will allow (and even an absence of empirical studies on the etiology and pathophysiology of insomnia by type and subtype) for the accumulation of enough data to conduct the meta- and mega-analyses that can address such issues.
References [1] Palagini L, Biber K, Riemann D. The genetics of insomnia-evidence for epigenetic mechanisms? Sleep Med Rev 2014;18(3):205e15. [2] Elder G, Wetherell MA, Barclay NL, Ellis JG. The cortisol awakening responsedapplications and implications for sleep medicine. Sleep Med Rev 2014;18(3):195e204. [3] Harvey CJ, Gehrman P, Espie CA. Who is predisposed to insomnia: a review of familial aggregation, stress-reactivity, personality and coping style. Sleep Med Rev 2014;18(3):217e27. [4] Spielman AJ, Caruso LS, Glovinsky PB. A behavioral-perspective on insomnia treatment. Psychiatr Clin North Am 1987;10:541e53. [5] Baglioni C, Regen W, Teghen A, Spiegelhalder K, Feige B, Nissen C, et al. Sleep changes in the disorder of insomnia. A meta-analysis of
Guest editorial / Sleep Medicine Reviews 18 (2014) 191e193 polysomnographic studies. Sleep Med Rev; 2014. http://dx.doi.org/10.1016/ j.smrv.2013.04.001. [6] Vgontzas AN, Bixler EO, Lin HM, Prolo P, Mastorakos G, Vela-Bueno A, et al. Chronic insomnia is associated with nyctohemeral activation of the hypothalamic-pituitary-adrenal axis: clinical implications. J Clin Endocrinol Metab 2001;86:3787e94. [7] Tuke DH. A dictionary of psychological medicine, vol. 1. London: JA Churchill; 1892. p. 61.
Michael L. Perlis*, Charles B. Corbitt Behavioral Sleep Medicine Program, Department of Psychiatry, University of Pennsylvania, USA
193
Center for Sleep and Circadian Neurobiology, University of Pennsylvania, USA Jacqueline D. Kloss Department of Psychology, Drexel University, USA * Corresponding
author. Behavioral Sleep Medicine Program, Department of Psychiatry, University of Pennsylvania, USA. E-mail address: [email protected] (M.L. Perlis). 22 January 2014 Available online 2 February 2014
Contents lists available at ScienceDirect
Sleep Medicine Reviews journal homepage: www.elsevier.com/locate/smrv
GUEST EDITORIAL
Insomnia research: 3Ps and beyond Introduction The four studies featured in this volume stand on the forefront of a new generation of insomnia research. Three of the four articles [1e3] bridge the past, present, and future of insomnia research by invoking the 3P model. This model, otherwise known as the Spielman Model [4], is a time-honored framework that has been exceptionally useful for the conceptualization of factors that contribute to the development and maintenance of insomnia disorder. Both of the genetics papers [1,3] focus on the predisposition for insomnia (both in terms of behavioral and molecular genetics). One of the two genetics papers [1] raises the interesting possibility that the perpetuation of insomnia may be ascribable to not only behavioral factors (e.g., sleep extension and the mismatch between sleep ability and sleep opportunity), but also to the epigenetic alteration of biologic processes. In addition to these foci, one paper [2] offers a description of an alternative approach to the assessment of hyperarousal and the measurement of cortisol. The other paper [5] offers a potential resolution to a long-standing issue: “Do patients with insomnia, as compared to good sleepers, exhibit sleep architectural abnormalities?”. The strengths of each of the four articles are highlighted below. This is followed by the suggestion that “before moving forward, it may be best to take a step back”. That is, our research efforts may be best served by (as the geneticists would say) making sure that the phenotype of interest is well characterized. Such an effort may be the only way to properly engage with, and learn from, the next generation of methods and measures. Brief review of article findings and positions Harvey and colleagues focus on the search for, and identification of, gene anomalies that may contribute to the pathophysiology of insomnia. They propose that a genetic vulnerability is conferred by 5HTTLPR, a serotonin transporter polymorphism that regulates serotonin in the synapse. Alterations in serotonin transport (reuptake from the synapse to the presynaptic component), are hypothesized to be related to disrupted sleep via an increase in stress reactivity. They further recognize the relevance of a genetic predisposition not only for new onset of insomnia, but also in the predisposition to the conditioned effects of insomnia. For example, after a few “bad nights” it is speculated that individuals with a genetic vulnerability will be less likely to recover compared to those without such vulnerability. Apart from the specifics of what may be learned from the genetic perspective, it should also be noted that the approach is valuable because it allows for a wholly different way of thinking, in that causal inferences can be made inductively from the genetic to the 1087-0792/$ e see front matter Ó 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.smrv.2014.01.003
biological to the psychological (5HTTLPR / serotonin dysregulation / stress reactivity / insomnia). This is in stark cont rast to the hypo-deductive approach (insomnia / hypera rousal / HPA activation / hypercortisolemia) which has been typical of insomnia research in the past. Further, this approach may yield novel perspectives on which biological processes and behaviors to target for the development of new therapies. For example, in the present case, 1) the 5HTTLPR finding suggests that modulation of serotonin may be as useful a strategy as the modulation of GABA with benzodiazepines and 2) psychological interventions for insomnia may do well to adopt therapies that specifically target stress reactivity. One such intervention, for which there is already good evidence, is the addition of mindfulness training to standard CBT-I. Palagini and colleagues’ paper is an ideal complement to the Harvey paper in that it not only focuses on the genetics of insomnia, but also extends that focus to include epigenetics. With respect to the genetics of insomnia, the authors provide estimates which indicate that 1) 35e73% of patients with insomnia report familial incidence with the mother being the most frequently affected relative, and 2) twin studies show heritability ranges from 21 to 55% depending on the composition of the sample (male/female; insomnia subtype [early, middle, late]; age of onset of insomnia). With respect to the epigenetics of insomnia, the authors first provide an explanation of the phenomenon and then go on to illustrate how it may be relevant to the predisposition toward, and perpetuation of, insomnia disorder. As defined in the paper, epigenetics refers to ‘the mechanisms by which long-term programming to genes may be altered and can provide for changes in gene function without changing the gene sequence’ (p. 5). This allows for the possibility that environment can modify the genetic predisposition for insomnia, both in terms of new onset and recurrence of the disorder. Further, epigenetic mechanisms may also account for intergenerational transmission. Unlike genetics, which has a more singular focus on risk and the permanent abnormalities that are associated with chronic illness, epigenetics has a focus on the possibility that environmental factors may contribute to both the development of, and recovery from, chronic illness. Elder and colleagues argue that the cortisol awakening response (CAR) holds the promise of providing an objective measure of a biological process that may serve to predispose, precipitate, and/or perpetuate insomnia. The CAR is the hypersecretion of (spike in) cortisol that occurs with the morning awakenings that terminate the sleep period. The spike begins with the final awakening and reaches its peak about 30 min later. The authors discuss best approaches to the assessment of the CAR. The method they recommend involves sampling immediately upon awakening and every
192
Guest editorial / Sleep Medicine Reviews 18 (2014) 191e193
15 min for an hour, on two consecutive days, while controlling for potential confounds (e.g., light, saliva contamination). This empirically-based protocol could be undertaken either in the laboratory or in ambulatory settings. Finally, the authors specifically recommend the study of the CAR in relation to factors associated with the 3P model of insomnia. For example, they suggest that the CAR could provide an index of the factors that: predispose individuals to insomnia (e.g., neuroticism, etc.); precipitate insomnia (e.g., life stress or perceived stress); and perpetuate insomnia (e.g., rumination or “caffeine use”). Baglioni and colleagues provide, by way of a meta-analysis, a summary of sleep continuity and architectural differences among subjects with insomnia (PIs) and good sleepers. These data represent the most definitive work within this arena. They provide a basis on which to judge PSG findings with regard to insomnia, both with respect to individual studies and group profiles. Significant findings were detected for both sleep continuity and sleep architecture, given data from a baseline night and from night 2 studies. With regard to sleep continuity, as a group, compared to good sleepers, subjects with primary insomnia had approximately: a 6 min longer sleep latency, six more awakenings per night, 22 min more wake after sleep onset, 23 min less total sleep time, and 7% lower sleep efficiency. With regard to sleep architecture, as a group, compared to good sleepers, subjects with primary insomnia had approximately 20 min less time in slow wave and 11 min less REM sleep. The authors suggest that these differences may have implications for cognitive (memory consolidation) and emotional (stress reactivity) processing. Further, it is suggested that the PSG findings are consistent with the hyperarousal model of insomnia, at least to the extent that the significant reductions in slow wave and REM sleep represent shallow sleep. Commentary Taken together, these four articles provide clarity with respect to basic PSG findings with insomnia, highlight the prospects for the neuroendocrine study of sleep continuity disturbance, and provide an overview of the prospects for genetic and epigenetic research for insomnia. These areas of research, along with other approaches (e.g., natural history studies, longitudinal assessments of nightto-night variability, stress reactivity protocols, etc.), and other methodologies (e.g., anatomical and functional imaging, imaging of neurotransmitter “levels”, etc.) suggest that insomnia research is poised for a new beginning. Prior to engaging in the next wave of research, it seems appropriate to revisit the problems that have potentially limited our findings to date. First, as genetic research gets underway and it becomes increasingly clear that there is a need for well characterized phenotypes, it is incumbent on us to revisit the long-standing issue regarding the heterogeneity of insomnia. As alluded to by two of the current authors [3,5], certain subtypes (and types) of insomnia may reflect not only different etiologies, but also inherently different pathophysiologies. Failing to appreciate this diversity may result in a failure to detect findings. This scenario is perhaps easiest to understand when considering insomnia subtypes. A classic example where sample homogeneity may have clarified findings comes with the work of Vgontzas and colleagues [6], with respect to cortisol hyper-secretion in patients with insomnia. In this classic study, cortisol was sampled on a 30 min basis and was shown to have a circadian distribution in both good sleepers and in patients with insomnia. The groups differed across the 24 h day in only one “zone”, during the sleep onset period (approximately 22:00 he 02:00 h). The authors do not interpret this effect as being attributable to sample composition. It is possible, however, that their sample was reliably populated by patients with sleep onset insomnia, and
that some large percentage of subjects did not have middle or late insomnia. This scenario may have allowed for the detection of what was otherwise a localized finding. Unlike the given scenario, it is not hard to imagine what typically occurs with non-specific recruitment. If it had been the case that the sample was equally populated by patients with only initial, or middle, or late insomnia; and if cortisol levels were otherwise near normal or potentially supra-normal (given within-night recovery); then, in smaller sample sizes with predictably large estimates of variability, one would expect to find no difference between good sleepers and the aggregate group of subjects with insomnia. Alternatively, significant differences might be found given a large enough sample, but these differences would likely be of small magnitude and appear trivial. A similar argument may be made for the insomnia types. Presumably, the diversification of so-called “primary insomnia” into specific diagnostic entities was undertaken because it was genuinely believed that not all insomnias are the same. That is, the insomnias fundamentally differ with respect to the nature, the developmental course of, and/or the factors that contribute to the chronicity, severity, and/or frequency of, the presenting complaint. While the notion that insomnia is not a singular entity is not new (with descriptions of insomnia types and subtypes dating back to at least the 1800s [7]), the framers of the ICSD-2 apparently believed in this so strongly that they put forward no less than four separate classifications for primary/chronic insomnia (psychophysiologic, paradoxical, idiopathic, and inadequate sleep hygiene insomnia). If these different forms of insomnia have different pathophysiologies, where some of the insomnias have some features while others do not, then the likelihood of observing significant differences between good sleepers and composite samples of patients with insomnia will again, be low. For example, if hypercortisolemia is a fundamental component of psychophysiologic insomnia, but not idiopathic insomnia (which may be related to inherent neurobiologic defects), then having a sample comprised of both types of subjects is not likely to separate between good sleepers and patients with insomnia on cortisol measures, except in large samples where the finding will appear small in magnitude. As said from the outset, it seems that our best path forward may be to take a step backwards. As we are poised to benefit and learn from new methods and measures, our best strategy moving forward may be to adopt the given classifications for type and subtype and to carefully profile all research samples in these terms. At minimum, this could simply be a few rows in a descriptive statistics table indicating what proportion of the sample met criteria for the various types and subtypes. At maximum, this could entail the provision of a supplemental data table providing means and standard deviations by group and subgroup for the dependent variables of interest. With time, such systematic tracking of subject characteristics and sample compositions will allow (and even an absence of empirical studies on the etiology and pathophysiology of insomnia by type and subtype) for the accumulation of enough data to conduct the meta- and mega-analyses that can address such issues.
References [1] Palagini L, Biber K, Riemann D. The genetics of insomnia-evidence for epigenetic mechanisms? Sleep Med Rev 2014;18(3):205e15. [2] Elder G, Wetherell MA, Barclay NL, Ellis JG. The cortisol awakening responsedapplications and implications for sleep medicine. Sleep Med Rev 2014;18(3):195e204. [3] Harvey CJ, Gehrman P, Espie CA. Who is predisposed to insomnia: a review of familial aggregation, stress-reactivity, personality and coping style. Sleep Med Rev 2014;18(3):217e27. [4] Spielman AJ, Caruso LS, Glovinsky PB. A behavioral-perspective on insomnia treatment. Psychiatr Clin North Am 1987;10:541e53. [5] Baglioni C, Regen W, Teghen A, Spiegelhalder K, Feige B, Nissen C, et al. Sleep changes in the disorder of insomnia. A meta-analysis of
Guest editorial / Sleep Medicine Reviews 18 (2014) 191e193 polysomnographic studies. Sleep Med Rev; 2014. http://dx.doi.org/10.1016/ j.smrv.2013.04.001. [6] Vgontzas AN, Bixler EO, Lin HM, Prolo P, Mastorakos G, Vela-Bueno A, et al. Chronic insomnia is associated with nyctohemeral activation of the hypothalamic-pituitary-adrenal axis: clinical implications. J Clin Endocrinol Metab 2001;86:3787e94. [7] Tuke DH. A dictionary of psychological medicine, vol. 1. London: JA Churchill; 1892. p. 61.
Michael L. Perlis*, Charles B. Corbitt Behavioral Sleep Medicine Program, Department of Psychiatry, University of Pennsylvania, USA
193
Center for Sleep and Circadian Neurobiology, University of Pennsylvania, USA Jacqueline D. Kloss Department of Psychology, Drexel University, USA * Corresponding
author. Behavioral Sleep Medicine Program, Department of Psychiatry, University of Pennsylvania, USA. E-mail address: [email protected] (M.L. Perlis). 22 January 2014 Available online 2 February 2014
