Self-injurious behavior (SIB) includes a wide range of behaviors, including some degree of physical or psychological self-damage. SIB includes non-suicidal self-injury (NSSI) and suicidal behavior (suicidal ideation, suicide plan, and suicide attempts) (1, 2). SIB is highly prevalent among adolescents and young adults. It is estimated that between 17.1% and 38.6% of adolescents experience lifetime SIB, and 7.83% repetitively (3). Individuals who engage in SIB are more likely to have mental disorders (4, 5). Noteworthy, SIB increases family health risks and the need for social services and medical attention (6). Although recent studies have yielded important insights into the prevalence, etiology, evaluation, and treatment of SIB (7, 8), the prediction and prevention of SIB remain challenging. A critical task in preventing SIB is to identify markers of risk for subsequent SIB to target prevention. Accordingly, the study of the role of biomarkers in SIB adolescents may have the potential value of identifying a high-risk group and helping to prevent suicide (9).

SIB leads to physical harm and pain; however, paradoxically, most self-injurers report feeling lower pain intensity (10, 11). Therefore, some researchers have suggested that altered pain processing may be a potential risk factor for SIB (2). Serum β-endorphin (BE) is thought to play an important role in SIB as BE regulates injury perception and analgesia in the body. The opioid deficiency hypothesis (12–14) argues that people who engage in SIB have low basal BE levels and raise them through repeated SIB. The addiction hypothesis proposes that people who exhibit SIB may be addicted to their own endogenous opioids (15). BE is known to induce feelings of euphoria and happiness by modulating dopamine neurons in parts of the brain associated with motivation and pleasure and influencing positive emotions to reward and reinforce behavior (16). SIB is thought to be an extreme response resulting from regulatory control in the face of negative emotions or stressful events (17). Thus, the release of BE activated by self-injury may help to reduce feelings of pain and increase pleasure (12, 18). This may lead to the addiction to SIB in vulnerable subjects (19). Indeed, our group has reported that adults addicted to suicidal behavior displayed higher serum BE levels than adult patients who are not addicted to suicidal behavior (20).

Previous studies have shown altered basal BE levels in adults with a history of NSSI (21). Unfortunately, very few studies targeted BE levels in adolescent self-injurers. This population is characterized by repetitive SIB, comorbidity, and a high risk of suicide (7). This study examines the relationship between psychopathology and BE levels with SIB. We hypothesized that: 1. Mental disorders mediate the relationship between BE levels and SIB; 2. Both the SIB and criteria of addiction to SIB correlate with BE levels, consistent with the addictive theory of suicide addiction.

Most included patients were born in Spain (n = 34), and the majority were Caucasians (n = 31). Eighty-seven percent (n = 34) were girls. The mean age of the sample was 15 years (14.92 ± 1.49). Fifteen percent (n = 6) of these adolescents were the only children in their families (see Table 1). The mean serum BE was 190.53 ± 74.83. Age positively correlated with BE levels (p = 0.03). The most frequent clinical diagnosis was depression (n = 25), followed by ADHD (n = 7). BE levels were lower in depressed patients (F = 3.96, df = 2, p = 0.03) and higher in patients diagnosed with ADHD and other diagnoses (see Figure 2A).

The mean onset age of NSSI was 13 years old (12.97 ± 1.55), Eighty-three percent (n = 33) of the patients also experienced SA, and the mean onset age of SA was 13 years old (13.67 ± 1.61). Seventy-four percent (n = 29) picked all PSS items with risky responses. No correlation between the PSS and SITBI with BE levels was detected. There was an association between the onset age of NSSI and the onset age of SA (p = 0.001, see Figure 1A), and the older the onset age of NSSI, the shorter the period between NSSI and the onset of SA (see Figure 1B). Based on this feature, we propose a classification: (1) the “Quick Transition” group: patients who began SA within 1 year after their first NSSI episode or who began their SIB career with a suicide attempt; and (2) the “Slow Transition” group: patients who passed more than 1 year between their first NSSI episode and their first SA. In addition, the “Unknown” group was comprised of patients who had begun NSSI <1 year before our data collection and had no SA. No difference was found between the quick transition group and the slow transition group (194.42 ± 83.77 vs. 187.96 ± 69.23, p = 0.81).

We also found a negative correlation between the S-PLE scores and BE levels in a simple linear regression model (β = −86.78, F = 7.98, p = 0.008, Adj. R² = 0.16). Half (n = 19) of the adolescents in the sample chose all answers that increased the risk of SA, except the question “If you have a partner, do you think he/she is unfaithful?”. In the DSM-IV-based addictive SIBs scale, eighty-seven percent (n = 33) of the participants fulfilled at least 6 out of the 11 criteria of SIB addiction; however, we found no statistically significant differences when comparing those addicted to SIBs to those not addicted (196.17 ± 74.19 vs. 147.73 ± 63.48, F = 1.82 p = 0.19; see Figure 2B). This difference may be statistically significant if the sample size increased to 123 individuals (assuming p = 0.05 and power = 0.8).

The present study reports several interesting findings regarding the association of serum BE values and sociodemographic and clinic characteristics in a sample of 39 adolescents hospitalized for SIB. The most relevant clinical finding was that the transition time from NSSI to SA is shorter with increasing age of onset at first NSSI. Regarding serum BE values, we found several interesting findings. First, the serum BE values were surprisingly high. Second, adolescents diagnosed with depression had lower serum BE values than those with other diagnoses, particularly ADHD. Third, the S-PLE global score correlated with serum BE values. Nevertheless, another interesting result even if it did not reach statistical significance, was that serum BE values increased with a higher degree of addiction to SIB.

Previous studies have reported that NSSI is an important factor for subsequent suicide behaviors (28, 29). In the present study, most patients who presented with NSSI also had at least one SA; a few patients had SA earlier than NSSI. Regarding the time sequence, the onset age of NSSI and SA were related. The earlier onset of NSSI is more likely to take a longer NSSI-SA transition time. Some studies suggested that this difference may be because those with earlier onset of NSSI lack the measures and knowledge to carry out SA earlier (30, 31). The BE did not differ between those who require longer NSSI-SA transition time (“Slow Transition”) and those who require less time (“Quick Transition”) to begin carrying out SA. Although more study is required, it seems that the transition between NSSI and SA does not lead to the variation of BE.

We found reduced BE levels in depressed patients, which is consistent with the findings of several previous studies (11, 32, 33). Nonetheless, some studies also reported opposite results (34, 35). One possible explanation is that the methodologies, sample gender and age, treatment medications, and screening for depression comorbidity possible conduce to different outcomes (36). Some studies showed that BE decreased in ADHD patients (37, 38). Our patients with ADHD as their primary diagnosis did not show the same decrease compared to other disorders. This may be because most ADHD patients in our sample were treated with medication, or because some medications could raise serum BE values (39).

In the analysis of other characteristics associated with SIB, the BE level significantly decreased as the S-PLE score increased. For adults, a score higher than 2.46 (1.70 has also been recommended as a cut-off point for increasing sensitivity) was interpreted as a high risk of suicidal attempt (40, 41). However, using S-PLE in the underage population still lacks study. Some questions are inappropriate for the adolescent population, for example, the question “If you had a partner, do you think he/she was unfaithful?” may sounds awkward for majority of adolescents. Therefore, the inverse relationship between S-PLE and BE serum levels should be interpreted cautiously. Finally, adolescents who met at least six criteria for SIB addiction had greater BE levels than those who met fewer criteria. However, for significant results, the expected sample should be >100, and patients in our study were mostly depressed. The effect of depression could overlap the difference in BE by different levels of addiction. Moreover, most adolescents in our sample had SIB for <3 years and the episodes were infrequent. A longitudinal study with a larger sample size may lead to significant results. In any case, a previous study of our team demonstrated a relationship between meeting criteria for an addiction to suicidal behavior and BE values.

This study aimed to discover potential variables related to variations in serum BE levels in a sample of adolescents hospitalized with SIB. Analysis of the characteristics of SIB and potential biomarkers may provide insight into the prevention of lifetime SIB. Our findings indicate that variations in BE levels are frequently related to multiple factors, although the limitations of the sample size prevented us to carry out some extra analyses that could reveal complex relationships. Future research using a larger sample and a longitudinal design with multiple measures is required to better comprehend the variations and peculiarities of BE serum values in adolescents. Moreover, we used interviews with patients and self-reports to collect questions related to SIB, which may be subject to biases such as concealment, exaggeration, and memory bias. Also, self-reported measures may reflect individual feelings and impressions rather than objective measures. Third, we lacked a control group which resulted in highly homogenized data and limited results. Fourth, our sample is gender-biased in the sense that most participants are female, which in turn makes our conclusions less generalizable and limits the identification of sex differences. However, as our study is still ongoing, a larger, balanced sample may lead to more significant results and allow more sophisticated analyses, such as Structural Equation Modeling analysis and machine learning-based models.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Comité de Ética de Investigación y Comité de Etica de Investigación con Medicamentos (CEIm). Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin. Written informed consent was obtained from the individual(s), and minor(s)' legal guardian/next of kin, for the publication of any potentially identifiable images or data included in this article.

HB-F contributed to conception and design of the study. PS-C, LM, PW, and HB-F recruited all patients and filled out the protocols. ED-N and EH-A managed β-endorphin analyses. MG-L, SR-G, and AS-L processed the samples at the biobank. PW organized the database. CL performed the statistical analysis. PW, CL, MB-F, MM-M, and HB-F wrote the first draft of the manuscript, contributed to manuscript revision, read, and approved the submitted version.

This study was funded by Alicia Koplowitz Foundation (Research Grant, 2020).

Author HB-F was employed by company Korian. In the last 24 months, HB-F received lecture fees from Shire. He is Principal Investigator (PI) of an iPFIS research contract (www.isciii.es; IFI16/00039) and co-PI of a MINECO research grant (RTI2018-101857- B-I00); recipient of: 1) a FIPSE Grant, and 2) an IDIPHIPSA intensification grant; involved in two clinical trials (MENSIA KOALA, NEWROFEED Study; ESKETSUI2002); CEO of Haglaia Solutions; member of the Advisory Board of ITA Salud Mental. Author Maria Rodrigo-Yanguas is the recipient of an iPFIS research contract (www.isciii.es; IFI16/00039). Author MM-M is the recipient of a CDTI grant (FEDER funded; IDI-20180701, file 00107278). The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.