Borderline personality disorder is a frequently studied condition that has its roots in childhood and adolescence, and is caused by multiple factors [3, 5–12]. Though it is widely accepted that some of these factors lay within family circumstances and parental rearing, this has mostly been evaluated retrospectively in community samples rather than clinical samples. In this study we examined perceived maternal rearing styles, and maternal psychopathology in a group of 101 adolescents with BPD features, 44 healthy controls, and their mothers. One of the strengths of our study is that the adolescents reported on perceived concurrent rearing by their parents, which makes the assessment less affected by recall bias. Also, this is the first study to report on referred adolescents with BPD features. Furthermore, to our knowledge, there is a paucity of research of parental psychopathology, both on Axis I and Axis II problems .
The main results of our study are as follows: (1) Adolescents with current elevated levels of BPD features report higher levels of maternal rejection, overprotection, and lower emotional warmth. (2) Mothers of adolescents with BPD features report more general psychopathology and cluster C personality symptoms, but no more cluster A and cluster B symptoms. (3) Three variables were the strongest predictors of BPD features in adolescents, namely the maternal rearing styles less emotional warmth and more overprotection, and more general psychopathology in mothers.
Contrary to expectations, we found no elevated levels of maternal cluster B personality traits in mothers of adolescents with BPD features. However, we did find higher levels of maternal cluster C traits in this group. Our findings differ from the study of Gunderson et al. , who found a 3- to 4- fold increased level of BPD in first-degree probands of BPD patients. Cheng et al.  found an increased risk for PD in students that were raised by a parent with personality pathology, but differences in (clusters of) parental or student PDs were not reported in their study. The relatively low level of cluster B personality traits in mothers of the clinical group is remarkable and might be due to methodological weaknesses in our study, such as the small sample size in the control group and the use of a self-report questionnaire for personality traits in mothers. However, our self-report measure on maternal BPD (PDQ-4) usually leads to higher rather than lower estimates of BPD. Next to the transgenerational transmission model,  another factor in the pathway to BPD is assumed to be dysfunctional parenting. Several studies have found ample evidence for this relationship [20, 23, 25, 26, 28]. This may be an explanation for the increased levels of cluster C traits in our sample: mothers with increased levels of cluster C (anxious, fearful) personality traits, may raise their children with more overprotection, and thus increase the risk of BPD in their offspring. Even more, in combination with general psychopathology (like anxious/depressive symptoms), those mothers may be unstable and unpredictable, and thus arouse instability in children that are already vulnerable for the development of BPD features.
Most (psychodynamic) theories on the development of BPD suggest that inappropriate parenting, like a lack of emotional warmth, high levels of parental criticism/regression, or overprotection, increase the risk of BPD symptoms. However, adolescents with BPD features may provoke these parenting behaviors, by their impulsive, instable, and dangerous behavior. These two causal directions may even reinforce each other . In addition, also other theories, including learning theories, have described the putative mechanisms involved in the development of BPD, and have included both parenting factors and offspring factors [13–17].
In our study, the maternal rearing styles emotional warmth and overprotection, together with increased general psychopathology in mothers, were associated with BPD features in adolescents. The model was able to classify 70% of the adolescents correctly (i.e. being assigned to the clinical or the control group) by using these three variables. Our findings are in line with the CIC study,  who found an association between aversive parental behavior and low parental affection, and BPD. However, they found no direct association between parental psychiatric disorders and increased risk for offspring PD. As in our study, Cheng et al.  found negative and conflicting parenting styles to be associated with the occurrence of personality disorder in general.
We found higher levels of emotional warmth in the control sample then in the clinical sample. Both parent and adolescent personality factors have been found to be relevant for influencing parenting behavior . High levels of extraversion and agreeableness in adolescents, and high levels of agreeableness in parents, results in better, positive parenting. It has been suggested that emotional stable parents are less anxious, and are therefore better able to handle problematic behavior in their adolescent children . Further exploration of moderators in the pathway to BPD is necessary, in order to develop interventions that aim at specific components of the disorder.
The contribution of parenting styles and maternal psychopathology is not unique to BPD features in youngsters . The same factors have been found to be associated with anxiety disorders  and depression . For example, overprotective parenting has also been found in a recent study with children with anxiety disorders (N = 190, age 7–13 years) . It is yet unclear what specific pathway leads to specific psychopathology.
There are, of course, some limitations. We used a cross-sectional design, so no causal interferences can be made. Though the sample size of the clinical group is large, the sample size of the control group is moderate. Further, our sample consisted almost only of girls, so generalization to a mixed population needs caution. On the other hand, this seems to reflect the general gender distribution among referred adolescents with BPD. Also, we did not report on severity of borderline symptoms. We used the self-report PDQ-4 to assess traits of personality disorders in mothers, an instrument that is known as sensitive, but also as non- specific. Moreover, we did not use a formal instrument to diagnose Axis I and Axis II disorders, not in mothers nor in adolescents (except for BPD symptoms in adolescents). It cannot be ruled out the results would have been different if we used a formal diagnostic interviews, e.g. the SCID-I  and SCID-II . Also, data on traumatic experiences is lacking. Another limitation is that we selected adolescents that already reported BPD symptoms. Their scores on the parental rearing scales may be affected by a response or attribution bias.
Another limitation is that we only reported on maternal rearing, leaving fathers out of consideration. Studies on BPD that included fathers in the assessment are scarce. The CIC- study reported on paternal rearing and paternal psychopathology, but the information was obtained through maternal interviews . Gureje et al.  did obtain information from both parents, but they did not report separately for fathers and mothers. However, it is likely that the role of fathers differs from the role of mothers. For example, it has been found that fathers influence the behavior of socially anxious children more than mothers . Bögels and Phares  propose a model for the role of paternal rearing in the development of anxiety disorders. They underline the importance of the role of fathers in the transition to the outer world, including encouragement of independence and appropriate risk-taking. The same may hold for adolescents with BPD and their fathers. Exploring the unique ways in which both fathers and mothers are involved in the etiology of BPD, is an interesting topic for future research.