Colin Harrington, MD

Self-injurious behavior (SIB) is intentional self-directed tissue injury inflicted without conscious intent to kill oneself (Simeon et al., 1992). A provisional nosology, based on the clinical phenomenology of self-injurious behavior, has been developed (Table 1) (Villalba and Harrington, 2000). However, this typology awaits etiological grounding. While a variety of neurotransmitter systems may be involved in the initiation and maintenance of repetitive self-injurious behavior (rSIB), most clinical studies to date have attended to the role of serotonin or endogenous opioids. This focus has emerged from a conceptualization of rSIB as a problem of impulse control (primarily mediated by serotonin) and/or as a maladaptive pain-related behavior (ultimately mediated by opioids) (Villalba and Harrington, 2000). Clinically significant forms of self-mutilation occur predominantly in mental retardation, pervasive developmental disorders, major psychotic disorders and cluster B personality disorders (Simeon et al., 1992). Psychodynamic formulations have suggested that self-inflicted injury may promote affect regulation, relieve anxiety, terminate dissociative experiences, discharge sexual arousal, support (dysfunctional) interpersonal relationships, generate euphoria, serve as self-punishment, establish ego/self boundaries and/or serve as a nonlethal alternative to suicide (Haines et al., 1995; Herpertz et al., 1997; Suyemoto, 1998). In this review, we address biological aspects of rSIB. 

Repetitive self-injurious behavior (except when induced by pain) may be fruitfully viewed as a developmental syndrome--highlighting the common denominator of its prevalence among Axis II disorders (both cluster B personality disorders and mental retardation). Self-injury may be triggered by depression, anxiety, psychosis, dissociation, pain, intoxication or incarceration; however, these factors rarely lead to rSIB in patients who are not predisposed to repetition prior to adult maturation. Two important developmental features of rSIB are 1) a sensitive dependence on critical periods (and/or critical events) that determine developmental trajectories and 2) the special role of learning/conditioning during an individual's formative years. 

Animal and human studies have linked early psychological trauma with subsequent rSIB. Psychological trauma (especially sexual abuse) and severe neglect may produce profoundly toxic epigenetic effects on neuropsychological development. Early social isolation of nonhuman primates frequently leads to rSIB, reduces dendritic branching in cortex and cerebellum, produces morphological changes in the striatum, alters hippocampal neuronal microstructure, and alters regional levels of norepinephrine (NE), dopamine (DA), serotonin (5-HT), substance P and leucine-enkephalin (Kraemer et al., 1997; Teicher, 2002). The hypothalamic-pituitary-adrenal (HPA) axis response to stress may also be blunted (Kraemer et al., 1997). 

Neural circuits that mediate the perception of pleasure and pain serve a "pedagogical" function by reinforcing adaptive behavior and extinguishing maladaptive behavior. When these pedagogical circuits develop aberrantly (as is presumed in rSIB), maladaptive characteristics may emerge, such as impulsivity, abnormal risk-taking, learning disabilities, attentional pathology and mood dysregulation.

The experience of pain is substantially altered in rSIB patients, compromising the most basic inhibition toward self-injury. Conditions in which known pain insensitivity accompany severe SIB include schizophrenia (Dworkin, 1994) and borderline personality disorder (Kemperman et al., 1997). The cooccurrence of experimentally verified pain insensitivity (Benedetti et al., 1999) and self-mutilation (Shua-Haim and Gross, 1997; Warnock et al., 1999) has also been observed in Alzheimer's disease (AD). 

Ascending neural pathways, diverging from the thalamus to discrete cortical regions, subserve different attributes of pain perception. Lateral thalamic nuclei project to somatosensory cortical loci important in localizing and assessing the intensity and duration of painful stimuli. The medial thalamic nuclei project to the anterior cingulate cortex (ACC), hypothalamus and amygdala, which mediate the perception of "pain affect" and "pain attention" (the aversive quality of pain). The pain insensitivity associated with AD may be attributed to deficits of attention mediated by the ACC, while sensory-discriminative capacity is preserved. The somatosensory cortex and thalamus are typically spared in AD, while conspicuous neuronal loss is common in the prefrontal and limbic cortices (Benedetti et al., 1999). The elevated pain tolerance of AD is associated with electroencephalographic slowing, suggesting a primary disturbance of vigilance. The pain insensitivity of AD may provide a model to explain the apparent pain insensitivity observed in other diagnoses displaying prominent attentional deficits (e.g., mental retardation, schizophrenia and dissociative disorders). Abnormal pain thresholds have been associated with hypnotizability and dissociative experience (Agargun et al., 1998).

An intrinsic pain inhibitory system can be activated in the presence or anticipation of pain. Stress may trigger the hypothalamus to release proopiomelanocortin (POMC), a prohormone that is subsequently cleaved to yield adrenocorticotropic hormone (ACTH), melanocyte-stimulating hormone (MSH) and ß-endorphin. The ACTH facilitates vigilance and stimulates the release of energy substrates to support the fight/flight response, while ß-endorphin provides relief from pain that may be encountered in fight or flight. In addition, b-endorphin may allow an animal to escape or fight despite injuries that would typically hamper mobility. This so-called "stress-induced analgesia" may account for the apparent analgesia experienced by wounded soldiers and athletes (Hilgard, 1976). Interestingly, superficial damage to skin produces both analgesia and an invigoration that facilitates fight/flight (Brodal, 1998).

Stress-induced analgesia may play a role in the apparent pain insensitivity of rSIB. Certain commonly observed aspects of rSIB support an endorphin hypothesis. These include: 1) psychic numbing associated with SIB episodes (opioid anesthesia), 2) escalation of the repetition and severity of rSIB (addiction) and 3) post-cessation dysphoria (withdrawal). Habitual SIB is associated with elevated mean plasma metenkephalin (Coid et al., 1983) as well as marked secretion of ß-endorphin immediately following SIB episodes (Sandman, 1990/1991). Sandman (1990/1991) discovered an uncoupling of ß-endorphin and ACTH secretion in some patients with chronic SIB. Since ACTH protects against opiate tolerance (Crofford and Casey, 1999), isolated ß-endorphin secretion predisposes tolerance and the drive to repeat an SIB. Moreover, without the vigilance-enhancing effect of ACTH, unopposed ß-endorphin binding in the ACC (Davis et al., 1997) may lead to hypovigilance and increased pain insensitivity. A similar uncoupling of ACTH and ß-endorphin has been reported in heroin addiction (Sandman, 1990/1991). A study of SIB in mentally retarded subjects found that a post-SIB dissociation of POMC peptides predicted a therapeutic response to naltrexone (ReVia) (Sandman, 1990/1991). Individuals that had the greatest post-SIB elevation of ß-endorphin showed the greatest improvement in SIB with naltrexone (Sandman, 1990/1991).

In some regards, cutaneous SIB resembles acupuncture, and substantial evidence supports an opioid mediation underlying the potent analgesia of acupuncture (Ulett et al., 1998; Wu et al., 1999). Acupuncture is typically administered to produce analgesia by penetrating the skin with sharp needles or by superficially burning it with lighted punks of 

 ("moxibustion"). Likewise, cutting and light burning are the most common types of cutaneous SIB. The distal forearm is a very effective site for inducing acupuncture analgesia (Wu et al., 1999) and tends to be a preferred site for cutaneous SIB. Brain imaging during the induction of acupuncture analgesia has demonstrated activation of the hypothalamus and its descending projections to the raphe nucleus and periaqueductal gray matter, and deactivation of rostral ACC, amygdala and hippocampus (Wu et al., 1999).

A final observation regarding the potential opioid involvement in SIB comes from the evidence pointing to a special role for opiates in mediating maternal-infant interaction. Mother's milk contains caseins that are digested into opioids. Moreover, nonnutritive suckling in humans and animals is extremely antinociceptive (Blass, 1997), which may account for the reinforcing feature of oral autostimulation (e.g., digit sucking). Human obstetric data suggest that a neonate's opiate responsivity is not fixed and may be "imprinted" (up- or downregulated) by environmental influence during a critical period (Jacobson et al., 1990). The opioid system's capacity to permanently imprint early influences may be important in accounting for the toxic epigenetic effects of child abuse and neglect that are found so prevalently among rSIB patients. A permanent deficit may induce perpetual opioid "craving" and consequential autostimulatory tendency. Rhesus monkeys reared in social isolation engage in a wide variety of self-injurious behavior, including head banging, self-biting, self-slapping and eye gouging. Some investigators have noted the parallel between the SIB observed in these monkeys deprived of maternal contact and the SIB observed in patients with severe cluster B personality disorders (many of whom have been abused or neglected during childhood) (Kraemer et al., 1997).

Substantial evidence inversely correlates peripheral and central markers of 5-HT function with impulsive, aggressive and suicidal behavior. Diminished 5-HT activity has been associated with impaired impulse control across a variety of conditions including depression, suicide, bulimia, cluster B personality disorders, alcohol abuse, repetitive self-injury of mental retardation and violent criminality (Coccaro and Kavoussi, 1997; Markowitz and Coccaro, 1995; O'Keane and Dinan, 1991; Virkkunen et al., 1994).

Human and nonhuman primates display stable trait-like interindividual variation in the serotonin system. A hypoactive serotonin system is associated with exploratory, risk-taking, impulsive and aggressive behavioral traits, whereas a hyperactive serotonin system is associated with a passive, docile, inhibited temperament.

Mehlman et al. (1994) demonstrated an inverse relationship between cerebralspinal fluid (CSF) 5-hydroxyindoleacetic acid (5-HIAA, a metabolite of serotonin) and violent, risk-taking behavior in free-ranging adolescent male rhesus monkeys. A related study found that the extent and frequency of fighting (and injuries) correlated inversely with CSF 5-HIAA (Higley et al., 1992). 

Asberg et al. (1976) first demonstrated that CSF 5-HIAA is reduced in depressed patients who attempted/completed suicide relative to depressed nonsuicidal controls. Further research in this area revealed a significant inverse relationship between CSF 5-HIAA and histories of impulsive aggression among patients with severe personality disorders (Asberg et al., 1976; Brown et al., 1982). Moreover, impulsive subjects who were historically suicidal had lower CSF 5-HIAA than their impulsive nonsuicidal controls. Work on autopsy material has reported decreased imipramine (Tofranil) binding in the frontal cortex of suicide completers compared to matched victims of accidental death (Coccaro and Astill, 1990; Stanley et al., 1982). A convergence of evidence suggests that, rather than directly regulating the commission of violent acts, serotonin modulates impulsivity as a specific dimensional trait (Linnoila and Virkkunen, 1992; Virkkunen et al., 1994; Virkkunen et al., 1987). The degree of SIB among patients with severe personality disorders correlates significantly with other measures of impulsivity, anger and somatic anxiety (Simeon et al., 1992).

Fenfluramine-induced prolactin changes may be more sensitive than baseline CSF 5-HIAA in indexing the serotonergic dysfunction associated with impulsive aggression (Coccaro et al., 1997). New et al. (1997) compared both prolactin and cortisol responses to fenfluramine challenge in 97 personality-disordered patients with and without histories of suicide attempts or SIB. Patients with histories of SIB or suicidal behavior had blunted prolactin and cortisol responses compared to controls. Subjects with a history of suicidal behavior and SIB had more pronounced prolactin and cortisol blunting than did subjects with histories of suicidal behavior alone--suggesting that central serotonin dysfunction is more sensitively related to SIB than "suicidality."

Treatment of rSIB has included trials of neuroleptics, lithium (Eskalith, Lithobid), anticonvulsant mood stabilizers, central α-2 agonists and ß-blockers. Recent interest has focused on serotonergic agents and opiate antagonists. Limbic leucotomy was reported to markedly reduce rSIB in four of five treatment-refractory patients (Price et al., 2001). Table 2 summarizes the pharmacologic treatment trials for rSIB. (See Pies and Popli [1995] and Villalba and Harrington [2000] for specific references.)

The following recommendations are provisional. More precise guidelines must await more controlled clinical studies.

Before initiating treatment, a behavioral/functional analysis of the SIB may uncover specific psychosocial precipitants.

 Determine the ratio of SIB subtypes presented by the patient. (For example, a patient may engage in the following constellation of SIB: 80% addictive-impulsive, 15% emblematic, 5% dissociative). Appreciating the subtype complexity of SIB that a single patient may present underscores the necessity for multiple tailored and appropriately weighted interventions. Psychotropic medication should be used to augment, not replace, behavioral therapy. Psychodynamic, behavioral and system-based treatments are important in the management of SIB (especially the Meaning Emphasis categories of SIB) (Hawton et al., 1998; Long and Miltenberger, 1998; Suyemoto, 1998).

If SIB occurs in the context of depression, mania or obsessive-compulsive behavior, treatment with an antidepressant, mood stabilizer or serotonergic agent is respectively recommended.

Determine whether psychosis, tics, seizures, akathisia or pain contribute to SIB and treat with an antipsychotic, anticonvulsant, tranquilizer or analgesic, respectively.

 Except in the case of tic-related SIB, atypical antipsychotics are preferred, given their serotonergic effects and better side-effect profiles. If seizures are implicated, carbamazepine (Tegretol) and divalproex sodium (Depakote) are specifically recommended. Central pain should be managed with agents that address its neurogenic origin.

In the remaining subtypes of rSIB, initiate treatment with a serotonergic agent. Trials of several different 5-HT agonists are warranted if the first agent selected is ineffective. 

If 5-HT agents are inadequate, consider substituting or adding naltrexone.

 (Nalmefene [Revex] may be a longer-acting alternative in the near future.) Recall that an opiate antagonist may require many months before producing its maximum benefit. A combination treatment of a 5-HT agonist and an opiate antagonist may have a theoretical advantage.

If all the above interventions are ineffective, trials of lithium (and other mood stabilizers), b-blockers (propranolol [Inderal], pindolol [Visken]),  α-agonists (clonidine [Catapres], guanfacine [Tenex]) and monoamine oxidase inhibitors are suggested.

Agargun MY, Tekeoglu I, Kara H et al. (1998), Hypnotizability, pain threshold, and dissociative experiences. Biol Psychiatry 44(1):69-71.
 

Asberg M, Trasksman L, Thoren P (1976), 5-HIAA in the cerebrospinal fluid. A biochemical suicide predictor? Arch Gen Psychiatry 33(10):1193-1197.
 

Benedetti F, Vighetti S, Ricco C et al. (1999), Pain threshold and tolerance in Alzheimer's disease. Pain 80(1-2):377-382.
 

Blass EM (1997), Milk-induced hypoalgesia in human newborns. Pediatrics 99(6):825-829.
 

Brodal P (1998), The Central Nervous System: Structure and Function. New York: Oxford University Press.
 

Brown GL, Ebert MH, Goyer PF et al. (1982), Aggression, suicide, and serotonin: relationships to CSF amine metabolites. Am J Psychiatry 139(6):741-746.
 

Coccaro EF, Astill JL (1990), Central serotonergic function in parasuicide. Prog Neuropsychopharmacol Biol Psychiatry 14(5): 663-674.
 

Coccaro EF, Kavoussi RJ (1997), Fluoxetine and impulsive aggressive behavior in personality-disordered subjects. Arch Gen Psychiatry 54(12):1081-1088.
 

Coccaro EF, Kavoussi RJ, Cooper TB, Hauger RL (1997), Central serotonin activity and aggression: inverse relationship with prolactin response to d-fenfluramine, but not CSF 5-HIAA concentration, in human subjects. Am J Psychiatry 154(10):1430-1435.
 

Coid J, Allolio B, Rees LH (1983), Raised plasma metenkephalin in patients who habitually mutilate themselves. Lancet 2(8349):545-546.
 

Crofford LJ, Casey KL (1999), Central modulation of pain perception. Rheum Dis Clin North Am 25(1):1-13.
 

Davis KD, Taylor SJ, Crawley AP et al. (1997), Functional MRI of pain- and attention-related activations in the human cingulate cortex. J Neuropsysiol 77(6):3370-3380.
 

Dworkin RH (1994), Pain insensitivity in schizophrenia: a neglected phenomenon and some implications. Schizophr Bull 20(2):235-248.
 

Haines J, Williams CL, Brain KL, Wilson GV (1995), The psychophysiology of self-mutilation. J Abnorm Psychol 104(3):471-489.
 

Hawton K, Arensman E, Townsend E et al. (1998), Deliberate self harm: systematic review of efficacy of psychosocial and pharmacological treatments in preventing repetition. BMJ 317(7156):441-417.
 

Herpertz S, Sass H, Favazza A (1997), Impulsivity in self-mutilative behavior: psychometric and biological findings. J Pychiatr Res 31(4):451-465.
 

Higley JD, Mehlman PT, Taub DM et al. (1992), Cerebrospinal fluid monoamine and adrenal correlates of aggression in free-ranging rhesus monkeys. [Erratum appears 1992;49(10):773.] Arch Gen Psychiatry 49(6):436-441 [see comment].
 

Hilgard ER (1976), Neodissociation theory of multiple cognitive systems. In: Consciousness and Self-Regulation, Schwartz GE, Shapiro D, eds. New York: Plenum Press.
 

Jacobson B, Nyberg K, Gronbladh L et al. (1990), Opiate addiction in adult offspring through possible imprinting after obstetric treatment. BMJ 301(6760):1067-1070.
 

Kemperman I, Russ MJ, Clark WC et al. (1997), Pain assessment in self-injurious patients with borderline personality disorder using signal detection theory. Psychiatry Res 70(3):175-183.
 

Kraemer GW, Schmidt DE, Ebert MH (1997), The behavioral neurobiology of self-injurious behavior in rhesus monkeys. Current concepts and relations to impulsive behavior in humans. Ann N Y Acad Sci 836:12-38.
 

Linnoila VM, Virkkunen M (1992), Aggression, suicidality, and serotonin. J Clin Psychiatry 53(suppl):46-51.
 

Long ES, Miltenberger RG (1998), A review of behavioral and pharmacological treatments for habit disorders in individuals with mental retardation. J Behav Ther Exp Psychiatry 29(2):143-156.
 

Markowitz PI, Coccaro EF (1995), Biological studies of impulsivity, aggression, and suicidal behavior. In: Impulsivity and Aggression, Hollander E, Stein DJ, eds. New York: John Wiley & Sons.
 

Mehlman PT, Higley JD, Faucher I et al. (1994), Low CSF 5-HIAA concentrations and severe aggression and impaired impulse control in nonhuman primates. Am J Psychiatry 151(10):1485-1491 [see comment].
 

New AS, Trestman RL, Mitropoulou V et al. (1997), Serotonergic function and self-injurious behavior in personality disorder patients. Psychiatry Res 69(1):17-26.
 

O'Keane V, Dinan TG (1991), Prolactin and cortisol responses to d-fenfluramine in major depression: evidence for diminished responsivity of central serotonergic function. Am J Psychiatry 148(8):1009-1015.
 

Pies R, Popli AP (1995), Self-injurious behavior: pathophysiology and implications for treatment. J Clin Psychiatry 56(12):580-588.
 

Price BH, Baral I, Cosgrove GR et al. (2001), Improvement in severe self-mutilation following limbic leucotomy: a series of 5 consecutive cases. J Clin Psychiatry 62(12):925-932.
 

Sandman CA (1990/1991), The opiate hypothesis in autism and self-injury. J Child Adolesc Psychopharmacology 1(3):237-248.
 

Shua-Haim JR, Gross JS (1997), Lesch-Nyhan syndrome in an Alzheimer's disease patient: a case report. J Am Geriatr Soc 45(8):1034.
 

Simeon D, Stanley B, Frances A et al. (1992), Self-mutilation in personality disorders: psychological and biological correlates. Am J Psychiatry 149(2):221-226.
 

Stanley M, Virgilio J, Gershon S (1982), Tritiated imipramine binding sites are decreased in the frontal cortex of suicides. Science216(4552):1337-1339.
 

Suyemoto KL (1998), The functions of self-mutilation. Clin Psychol Rev 18(5):531-554.
 

Teicher MH (2002), Scars that won't heal: the neurobiology of child abuse. Sci Am 286(3):68-75.
 

Articles

Repetitive self-injury can be one of the more difficult conditions to treat. What is the biochemical basis for self-injury and how can psychiatrists treat this condition?

Colin Harrington, MD

Articles

Repetitive Self-Injurious Behavior: The Emerging Potential of Psychotropic Intervention