COL1A1 and COL1A2 sequencing results in cohort of patients undergoing evaluation for potential child abuse

INTRODUCTION

Child abuse is a major public health concern that is the result of a complex sociopathology. Fractures in children are common and physical abuse is responsible for a small proportion of childhood fractures [Kemp et al., 2008]. A detailed review of the clinical history in conjunction with a comprehensive examination are the first steps in the evaluation of children who may have been victims of abuse [Jenny, 2006; Flaherty et al., 2014; Pereira, 2015]. Conditions that can result in bone fractures and/or periosteal elevation in pediatric patients include inherited conditions such as osteogenesis imperfecta (OI), hypophosphatasia, disorders of copper metabolism, some types of rickets, and Caffey disease [Marlowe et al., 2002; Bishop et al., 2007; Pandya et al., 2011; Singh Kocher and Dichtel, 2011; Flaherty et al., 2014; Metz et al., 2014]. The clinical recognition of an underlying bone fragility predisposition, such as OI, is often challenging in infants and toddlers because some of the cardinal signs (short stature, blue sclera, teeth anomalies, osteopenia, family history, and the presence of wormian bones on skull X-rays) may not have emerged or be found [Kocher and Shapiro, 1998; Marlowe et al., 2002; van Dijk et al., 2011; Flaherty et al., 2014].

With prevalence between 0.4/10,000 and 0.79/10,000, OI is the most common inherited syndrome that predisposes to skeletal fractures [Orioli et al., 1986; Stevenson et al., 2012]. It comprises a heterogeneous group of disorders with variable severity and is usually caused by pathogenic variants in the type I collagen genes, COL1A1 and COL1A2 [van Dijk et al., 2011]. Diagnostic tests include analysis of type I collagen synthesized by cultured fibroblasts and DNA sequence of the coding, and flanking splicing elements in both genes. The latter, now considered the preferred starting point, can identify the cause of OI in close to 95% of affected individuals and will miss those with rare large deletions or duplications in the collagen genes, and those with pathogenic variants in other genes [van Dijk et al., 2012, 2011; Pepin and Byers, 2015].

The incidence of OI among children with fractures that are evaluated in the context of child abuse has been estimated to be between 2–5% [Steiner et al., 1996; Marlowe et al., 2002; Pepin and Byers, 2015]. In this study, we sought to determine the indications, results, and clinical interpretation of molecular testing for OI sent to a reference laboratory in a group of patients evaluated for child abuse.

PATIENTS AND METHODS

After obtaining approval from the Arkansas Children's Hospital Institutional Review Board, a retrospective review was performed of all patients with bone fractures in whom physical abuse was suspected and molecular testing for OI was performed at a reference laboratory between July 2011 and September 2014. All available clinical data were abstracted from the medical records including: age at testing, gender, genetic evaluation, family history, multidisciplinary evaluation for possible child maltreatment, and radiographic and metabolic studies. Family history was determined to be positive if first-degree relatives were thought to have OI on clinical grounds alone or if the diagnosis was confirmed by genetic tests. All patients were evaluated by the Team for Children at Risk (TCAR), a team dedicated to provide inpatient and outpatient consultations for maltreatment-related concerns including evaluation of injuries as well as for multiple forms of neglect. As part of the evaluation, all children are routinely screened for a family history of unexplained fractures (including OI), and for the presence of other findings on physical exam suggestive of OI such as blue sclerae, triangular-shaped face, and skeletal anomalies.

For the purpose of this study, the indication for OI molecular testing was recorded as “maltreatment concerns” if there were no clinical findings to suggest OI. Alternatively, if a clinical feature suggestive of OI was present (blue sclerae, osteopenia, wormian bones, triangular-shaped face, a positive family history of OI, or short stature) this was recorded as the indication for testing. Similarly, the final diagnosis or disposition was determined by a physician from the children at risk team (R.C.) once the assessment of the psychosocial environment, the mechanism of injury reported, family history, and clinical and laboratory results was completed. The non-mutually exclusive possibilities included: physical abuse, accidental injury, OI, other inherited metabolic predisposition syndromes, rickets, or indeterminate.

RESULTS

Molecular testing for OI was performed in a total of 43 patients (72% male). The median age for testing and number of fractures were 5 months (range = 3 weeks to 62 months) and six fractures (range = 1–40), respectively (Table I). Soft tissue injuries such as lacerations and bruises were documented in 37% and none (0/43) had wormian bones detected by skeletal survey. Other clinical and laboratory evaluations failed to identify an explanation for fractures. For the whole population, molecular testing for OI was performed for maltreatment concerns in the absence of other clinical indicators in 35 cases (81%), while the final diagnosis was that of only physical abuse in 33 cases (77%) with no patients found to have an inherited metabolic predisposition or rickets (Table II). For those patients that received the final diagnosis of physical abuse, rib fractures, and classic metaphyseal lesions of the femur were the most common.

None of the patients tested in which child abuse was at the top of the differential diagnosis after clinical assessment had pathogenic variants found in the COL1A1 or COL1A2 genes (0/35). Pathogenic variants in type I collagen genes were identified in two individuals (5%), both with clinical signs that suggested the diagnosis (Table II). Patient #4 was a 3 month old girl who was admitted to the hospital with fractures of the femoral metaphysis, tibial metaphysis, scapula and multiple ribs, the lack of trauma history, and no family history of OI. The patient had osteopenia, blue sclera, and short length at the time of her initial assessment. Sequencing analysis of the type I collagen genes, COL1A1 and COL1A2, identified a previously unreported COL1A2 variant (c.1199G > C, p.Gly400Ala, Gly310Ala in the triple helical domain) that was predicted to be pathogenic because it substituted a canonical glycine in the Gly-Xaa-Yaa repeating triple of the triple helical domain. Individuals, or one of more family members, identified to date who have such substitutions have an OI phenotype. In this instance, the variant was inherited from her phenotypically normal mother. At 2 years of age, the patient has not had additional fractures and continues to be followed by genetics. Given the pattern of fractures and the lack of a plausible explanation of injury, the final assessment from the TCAR was that of combined physical abuse and OI. In patient #8, OI was clinically suspected at 7 months of age given a family history of OI and the presence of blue sclera. However, because she was found to have only a skull fractures at presentation, a full evaluation by the TCAR was recommended, and the final diagnosis following that evaluation was of accidental injury. A previously reported COL1A1 pathogenic variant (c.796G > A, p.Gly266Arg, Gly88Arg in the triple helical domain) was found but parental studies were not performed.

Three patients had variants of unknown significance (VUS) (7%). Patient #12, a 2 year old who had had four fractures during his lifetime, was found to have a COL1A1 intronic change (c.104-13_−12CC > AA) that creates a new splice acceptor site that might be used in some transcripts. The patient also had blue sclerae but no other significant findings on exam or relevant family history. Subsequent familial or functional studies were not conducted after the family moved out of state. The suspicion for child abuse for this patient was low. Patients #42 and #43 each had a VUS that was likely benign as predicted by in silico tools. The former was not found to have any evidence of abuse while the latter was diagnosed with child physical abuse.

DISCUSSION

Currently, there are no consensus guidelines to aid the clinicians who are trying to decide if genetic testing for OI is needed in the context of fractures or when there is concern that child abuse may be the cause. Some authors emphasize the utility of clinical recognition of the features of OI, rather than universal laboratory testing while others suggest that testing helps with potential legal and psychosocial implications when the clinical distinction between OI and child abuse is not apparent [Kocher and Shapiro, 1998; Byers et al., 2006; Bishop et al., 2007]. In the recent paper by Pepin and Byers, [2015] the authors recommended that when testing for OI sequencing of the COL1A1, COL1A2, and IFITM5 genes, in addition to deletion/ duplication analysis, should be conducted first. Despite this recommendation, the question as to what criteria to use to determine who needs to be tested in the setting of potential child abuse still remains.

In this study, we presented a unique perspective from the clinical side, and reviewed the indications and results of OI molecular genetic testing as well as the final diagnosis, as determined by the multidisciplinary evaluation team, in a cohort of children with fractures that was evaluated for child abuse. None of the patients tested for OI when maltreatment was suspected were identified to have pathogenic variants. Conversely, the two cases with confirmed OI had some clinical signs that were easily recognizable and that led to the initial suspicion of OI.

Published recommendations for non-genetic testing in children who are evaluated for child abuse include radiological skeletal survey, computed tomography of the head, complete blood cell count, serum calcium, phosphorus and alkaline phosphatase concentrations, and serum 25-hydroxy-vitamin D concentration [Jenny, 2006; Kellogg, 2007; Johnson, 2009; Flaherty et al., 2014]. No other inherited metabolic bone fragility predisposition was diagnosed in this study. Wormian bones are reported in up to 78–90% of patients with OI type IV and not seen in healthy controls [Semler et al., 2010; Bellary et al., 2013]. In this cohort, none of the 43 patients evaluated for child abuse were found to have Wormian bones. That the two patients in this cohort with pathogenic variants did not have them is an important reminder of the utility of this finding when present but the limited predictive value in this setting when absent.

Although any type or site of fracture can be caused by child abuse or OI, some fractures, such as the classic metaphyseal lesion and posterior rib fractures, are more frequently found in abused children [Byers et al., 2006; Jenny, 2006; Bishop et al., 2007; Kemp et al., 2008; Johnson, 2009; Dwek, 2011; Greeley et al., 2013; Kodner and Wetherton, 2013; Flaherty et al., 2014; Pereira, 2015]. Given the small numbers and the fact that one patient with OI also was diagnosed with physical abuse, no direct comparison of the location of the fractures was possible. Lastly, as illustrated by one of our cases, the presence of a pathogenic variant consistent with OI does not exclude the possibility of physical abuse. For this patient, even though there were some clinical findings suggestive of OI, the pattern of fractures and the lack of a plausible explanation of the injuries led to the final diagnosis of child abuse. A careful review of the medical history and a detailed clinical evaluation are paramount to help make the distinction in these cases.

The present study is retrospective, limited to those instances in which genetic testing was performed and so is subject to the limitations intrinsic to such an investigation. Despite the clinical suspicion of OI in some cases that justified the molecular testing, the choice about which individuals to test was in many instances subjective or influenced by psychosocial and legal factors, and ultimately the motivation to conduct this project. The population tested for OI in this study represents a selective sample and ignores all the patients who had fractures and were diagnosed with child abuse and did not get tested. Therefore, while the 5% frequency of pathogenic collagen variants consistent with OI found in this study is similar to what was previously reported, this is most likely an overrepresentation of the true percentage of children with fractures where OI needs to be considered and then diagnosed [Marlowe et al., 2002; Pepin and Byers, 2015]. Because of the previous recommendation to stop after COL1A1 and COL1A2 sequencing unless there is a clear clinical diagnosis of OI or strong suggestion of consanguinity or recessive inheritance, less common types of OI caused by pathogenic variants in other genes were not assessed [van Dijk et al., 2012; Pepin and Byers, 2015]. Similarly, no biochemical fibroblast protein studies were conducted to complement the DNA sequencing.

While molecular testing for OI is most likely conducted in the inpatient setting, and therefore, “tagged” to the total inpatient bill, testing for COL1A1 and COL1A2 (and now IFITM5) with deletion/duplication analysis could reach over $2,000 per patient. In general, it is our experience that molecular testing for OI is routinely paid by the child's health insurance whether public or private, since the test is usually requested as part of the child's medical evaluation for multiple and/or unusual fractures. The additive cost burden to the health care system from OI molecular testing in the setting of potential child abuse needs to be weighed against the cost of potentially permitting further abuse to occur and the lifetime cost of child maltreatment with the resulting lost productivity, and increased lifetime health care costs. It is imperative to continue to work on defining what criteria to apply before deciding if testing for OI is needed in the setting of potential child abuse.

In summary, the frequency of causative DNA variants in this set of patients evaluated for child abuse was low. Routine testing for OI for maltreatment concerns did not identify any pathogenic variants while careful clinical evaluation was sufficient to clinically suspect OI. Because of the possibility of concurrent child abuse, evaluation of the abnormal results, and the clinical context is needed to reach the final diagnosis. Given the results of this study, we suggest considering molecular testing for OI and a genetic evaluation in the setting of child abuse only in the presence of blue sclera, osteopenia, and/or a positive family history.