Introduction

Liver biopsy is a beneficial test for the evaluation of children with liver disorders [1, 2]. Liver biopsy and histology, although invasive, has a prominent role in the definitive diagnosis and staging of indeterminant liver disease [3]. This may change as diagnostic measures improve [3, 4]. However, a high-quality liver biopsy provides information regarding etiological diagnosis, disease stage and progression as well as response to treatment at an extensive level that few other diagnosis measures can consistently provide [5, 6]. Therefore, obtaining adequate sampling while minimizing complications is critical for subsequent therapeutic management and prognostic implications.

Percutaneous liver biopsy (PLB) is the standard method for liver tissue sampling [7, 8]. This approach, however, introduces an increased risk of hemorrhagic complications, especially in the presence of coagulopathy, thrombocytopenia or ascites [9]. For this reason, PLB is contraindicated in an estimated incidence of 12% in children [10] requiring liver biopsy.

The transjugular liver biopsy (TJLB), first described by Hanafee and Weiner in 1967, is a method of sampling used to obtain liver tissue via the hepatic veins by puncturing the internal jugular veins [11]. This approach has gained growing recognition in adult patients with contraindications to PLB. As this technique has been modified, the spectrum of patients eligible for TJLB has expanded to include younger and smaller patients. TJLB has the advantage of measuring the hepatic venous pressure gradient simultaneously, but the quality of the TJLB specimens has been questioned due to smaller and more fragmented tissue in comparison to PLB [12, 13]. The use of radiological guidance (ultrasound and fluoroscopy), for example, has enabled a better visualization of the needle position, allowing the targeting of smaller liver lobes, and increasing operator confidence in smaller, younger patients and liver transplant patients with variable anastomosis [14]. This has lowered the mean number of passes required to catheterization the vein (from 4.2 to 1.5), reducing the frequency of neck complications (from 2.8 to 1.9%), and lowering the rate technical failures (from 0.8 to 0.2%), of which 26% are attributed to failure to cannulate the IJV [15, 16].

However, the data concerning TJLB in pediatric patients remains limited [17,18,19]. The first reported pediatric series was conducted in 1992 by Furuya et al., nearly 25 years after TJLB was first introduced [18]. In the following years, few papers have reported their experience with large numbers of pediatric patients. Moreover, studies of pediatric TJLB have observed smaller biopsy samples, more fragmented core biopsy samples, and less average number of portal tracts compared to adult TJLBs [10, 20, 21]. In contrast to adults, there is an increased intraprocedural technical difficulties due to lower body weight, smaller patient livers and horizontal hepatic veins in younger patients which may increase the risk of capsular perforation [15]. These factors can, in turn, affect the quality of the samples and outcomes of the technique [17, 22]. Other drawbacks to TJLB include radiation exposure and the relatively high cost of the procedure [10, 23]. Choice of liver biopsy technique therefore remains with the personal experience and judgement of the clinician on the basis of the age-risk profile of the individual patient [24].

There is a need, therefore, to determine whether TJLB adds any diagnostic information to the management of pediatric liver disease in patients at high risk of bleeding, and whether these gains outweigh the greater technical difficulty in obtaining histologically adequate samples. To our knowledge, no meta-analysis has investigated the use of TJLB in the pediatric population. In this meta-analysis, the aim is to evaluate the safety and efficacy of TJLB performed in pediatric patients by evaluating the rate of technical success, biopsy adequacy, and complications.

Methods

Search Strategy and Data Sources

A comprehensive search of several databases from inception to August 2022 was conducted in compliance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [25]. The databases included Ovid Cochrane Library, Medline, Embase, Epub, Scopus, PMC Preprints, and ClinicalTrials.Gov. The search strategy was designed and conducted by an experienced librarian with input from the study’s principal investigator. Controlled vocabulary supplemented with keywords was used to search for transjugular liver biopsy and pediatric. The actual strategy listing all search terms used and how they are combined is available in Fig. E1. This review was registered prospectively with PROSPERO (CRD42022354421).

Eligibility Criteria and Quality Assessment

Eligible studies must have met all the following participant inclusion criteria: (1) < 18 years old; (2) undergoing transjugular liver biopsy; and (3) reporting the outcome of adequacy of histopathological sampling, technical success rate, and postprocedural complications. Case reports, case series, abstracts, conference abstracts, and articles that were not reported in English were excluded from the study. This meta-analysis did not exclude studies on the basis of sample size. Article screening and data extraction were conducted by four independent assessors (KS, NY, AS, FA). Any disagreements were adjudicated by SM and discussed with co-authors as necessary. The quality of each study was independently evaluated by two authors (KS and FA) using the Newcastle–Ottawa Scale [26]. Any discrepancies was discussed by two independent assessors, with disagreements addressed via an adjudicator (SM). Results of the quality assessment of all included studies are shown in Fig. 2.

Statistical Analysis

The pooled means and proportions of the data were analyzed using a random-effects, generic inverse variance method of DerSimonian and Laird, which assigns the weight of each study based on its variance [27]. The heterogeneity of effect size estimates across the studies was quantified using the Q statistic and I2 (P < 0.10 was considered significant). A value of I2 of 0–25% indicates insignificant statistical heterogeneity, 26–50% low heterogeneity, 51–100% high heterogeneity [28]. Furthermore, a leave-one-out sensitivity analysis was conducted to assess each study’s influence on the pooled estimate by omitting one study at a time and recalculating the combined estimates for the remaining studies. Publication bias was assessed using a funnel plot [29]. If mean and standard deviation (SD) were unavailable, the median was converted to mean using the formulas from the Cochrane Handbook for Systematic Reviews of Interventions [28]. Data analysis was performed using Open Meta analyst software (CEBM, Brown University, Providence, Rhode Island, USA).

Results

Study Selection and Patient Characteristics

The initial literature search of electronic databases resulted in a total of 921 studies. After removing duplicates, 919 articles underwent screening for inclusion and exclusion criteria, with 35 full-texts evaluated for eligibility. Eight studies [14, 17, 23, 30,31,32,33,34] involving 361 patients met the criteria and were included in this quantitative meta-analysis. All the eligible studies were retrospective studies. The date of publication ranged between 2003 and 2021. Seven studies were based on single-center data, and one was a multicenter study [34]. The age of patients was reported in four studies [23, 30,31,32] and ranged from 56 to 176 months. Of the total 361 patients, 149 (41.27%) were female. The PRISMA flow chart describing the study selection process is depicted in Fig. 1. All included studies followed the principles detailed in the Declaration of Helsinki. The baseline characteristics of the included studies are comprehensively described in Table 1.

Fig. 1
figure 1

PRISMA flow diagram

Table 1 Baseline characteristics

Risk of Bias and Quality Assessment

Results of the quality assessment of all included studies are shown in Fig. 2. The Newcastle–Ottawa Quality Assessment scale was used to evaluate the risk of bias and quality of included studies for this meta-analysis. As such, studies were categorized into good, fair, and poor quality according to scores attained in the selection, comparability, and outcome/exposure domains. Studies were determined to be of moderate quality. The leave-one-out sensitivity analysis conducted on the technical success rate and adequacy of histological sampling demonstrated no difference between with and without the omitted study.

Fig. 2
figure 2

Newcastle–Ottawa Quality Assessment Scale for included studies. N/A not appropriate

Clinical Characteristics

Among eight studies, a total of 361 patients underwent TJLB. Three studies [23, 30, 31] reported the mean patient weight, calculated to be 40.46 ± 13.76 kg. The mean INR was reported in four studies [23, 30,31,32] at 1.81 ± 0.67, and the mean platelet count was reported in four studies [23, 30,31,32] at 88,621.74 ± 82,970.63 cells/mm3. Among seven reported studies, 89 (24.65%) patients had ascites [14, 17, 30, 31, 33, 34]. Upon clinical assessment, it was found that 27 patients (7.46%) had an initial diagnosis of portal hypertension, 36 (9.94%) had biliary atresia, 9 (2.49%) had autoimmune hepatitis, 54 (14.91%) had a fulminant hepatic failure, 3 (0.83%) had liver cirrhosis, 9 (2.49%) had Wilson’s disease, 13 (3.59%) had graft rejection, 13 (3.59%) had acute/subacute liver failure, 93 (25.69%) had chronic liver disease, and 105 (29%) had other diagnoses. The baseline clinical characteristics of the included studies are comprehensively described in Table 2.

Table 2 Clinical characteristics

Periprocedural/Technical Characteristics

There was a total number of 374 tissue biopsies performed from a total of 361 patients. For 311 patients (86.15%), the needles used during the procedure were 18/19- gauge quick-core needles. All eight studies used needles manufactured by Cook Medical. A total of 189 patients were administered general anesthesia (52.35%), 102 received local anesthesia (28.25%), and IV sedation was administered to 27 patients (7.47%). Five studies [14, 17, 23, 30, 31] used ultrasound guidance to obtain jugular venous access. Five studies performed the procedure under fluoroscopic guidance [14, 17, 31, 33, 34], and one study [14] used it in combination with trans-abdominal ultrasonography. The level of expertise of the operator was given in four studies [23, 32,33,34]. The corresponding data is illustrated in Table 3.

Table 3 Perioperative/technical characteristics

Technical Success Rate

In four studies, the procedure was considered technically successful if the operator considered an adequate tissue sample had been obtained [23, 31, 33, 34]. Two studies defined technical success as obtaining  a histopathologically adequate sample as specified by the number of visible portal tracts and total length of specimen [17, 30]. The remaining two studies did not define technical success [14, 33]. All eight studies reported the technical success rate from a total of 374 biopsies (Fig. 3). The pooled rate of technical success was evaluated to be 99.1% (95% CI 98.2%, 100.1%; I2 = 0%; P = 0.946). Of these, there were three procedural failures reported in six studies [14, 17, 23, 30, 31, 34]. The hepatic vein could not be cannulated due to technical difficulties in one case [23], and lack of muscle relaxant causing respiratory motion in the second case [14]. The procedure was abandoned in the third case due to hemodynamic instability [30]. The weighted median number of passes was 2 ± 1, as reported in three studies [23, 30, 31]. The corresponding data is detailed in Table 3.

Fig. 3
figure 3

Pooled estimate of technical success rates of TJLB. TJLB transjugular liver biopsy

Specimen Adequacy

All eight studies reported their histopathological adequacy of tissue samples from a total of 374 specimens (Fig. 4). As such, the pooled rate of histopathological adequacy of tissue samples was evaluated to be 97.5% (95% CI 95.4%, 99.5%; I2 = 27.66%; P = 0.208). Of these specimen tissues, two studies reported their average total length of sample, which was 14.65 ± 5.54 mm [23, 31].

Fig. 4
figure 4

Pooled estimate of adequacy of sampling rates of TJLB. TJLB transjugular liver biopsy

Postprocedural Characteristics

As summarized in Table 2, there was a total of 362 initial diagnoses made from clinical assessment. Following the procedures, 19 new diagnoses were established while 150 were confirmed.

Complications and Adverse Outcomes

In this meta-analysis, postprocedural complications were reported in six studies [14, 23, 30, 32,33,34]. Five of the eight studies clearly divided complications into major and minor [14, 17, 23, 30, 31]. Two of the included studies used the criteria for classification of procedure-related adverse events detailed by the Society of Interventional Radiology [23, 34]. One study [33] categorized the complications according to the CIRSE classification system. Finally, Oshrine et al. did not use a specific classification system, however they defined a complication as “an adverse event that occurred as a direct result of the liver biopsy" [32]. There was a total of 49 complications reported from the included studies. Of the most common, 19 patients (38.78%) experienced minor bleeding from the entry site, 6 (12.24%) had pyrexia for less than 24 h, 5 (10.2%) required red blood cell (RBC) transfusion, 4 (8.16%) had supraventricular tachycardia, and 4 (8.16%) required analgesia for pain. A total of 13 deaths were recorded, due to the procedure or postprocedural complications or as a result of their primary disease [14, 23, 30, 32, 33]. There was a single procedure-related death. A 3.5-year-old girl developed ventricular tachycardia when the guidewire was withdrawn, which progressed to cardiac arrest and subsequently death [14]. The guidewire was not thought to have been the cause, as autopsy reported no blood in the peritoneal cavity. A focal contraction band necrosis localized to the ventricles was designated as the origin of the arrhythmia. Another death was due to haemobilia post-procedure causing worsening hepatic encephalopathy and need for liver transplant. The patient died 34 days later due to transplant complications [23]. The other 11 patients died due to complications of their primary disease [14]. The data concerning postprocedural outcomes are summarized in Table 4.

Table 4 Post-operative characteristics

Discussion

Transjugular liver biopsy (TJLB) is well regarded in the literature for its safety, especially in patients with marked coagulopathies and ascites where this modality is indicated over the percutaneous approach. Technically, TJLB utilizes a transvenous route of access which subsequently confines bleeding to the venous system and avoids complications that are traditionally encountered in the intraperitoneal cavity [20]. However, the quality of TJLB samples is often questioned in the literature [15], especially in children [10]. The primary aim of this meta-analysis was to investigate the safety and efficacy of TJLB in patients with liver disease, specifically in the pediatric population as no previous meta-analysis has explored this population subset. Following the inclusion of eight studies within this meta-analysis, TJLB was found to exhibit high rates of technical success and adequacy of histopathological sampling. Furthermore, six of the eight studies reported postprocedural complications. Overall, the findings exhibited from this study may provide insight for physicians when selecting the suitable modality for conducting liver biopsies in selected pediatric populations.

Assessing the criteria for adequacy of tissue sampling requires specific consideration of the total length of the tissue sample and the number of visible complete portal tracts. Recording of a total length that is 15 mm or greater, obtained via multiple passes, is often considered to be the cut-off value appropriate for histopathological analysis [20]. The most recent adult meta-analysis by McCarty et al. reported achieving an average total sampling length of 15 mm or greater [35]. The weighted mean average length of tissue samples was 14.65 ± 5.54 mm calculated from two studies in this meta-analysis, which is close to the aforementioned value. One reason for this slightly lower value can be attributed to the fact that TJLB in pediatric patients are limited by smaller, more fragmented tissue samples, which therefore consist of fewer portal tracts [10, 20]. As a result, this can influence the adequacy of tissue sampling as per the above criteria. Similarly, the median number of passes calculated in our study was 2, which was slightly lower than Kalambokis et al., who in their meta-analysis reported 3 (average of 2.7) [15]. Despite this, the pooled rate of histopathological adequacy of sampling was calculated at 97.5% with four studies reporting 100% histopathological adequacy [17, 31,32,33]. Miraglia et al. suggested that the high rate of adequate sampling could be associated with the patient’s initial diagnosis, in which the majority of cases were “non-cirrhotic” [31]. Interestingly, this was not compromised in the other three studies [17, 32, 33], which made no mention of any patients with an initial diagnosis of liver cirrhosis. Instead, chronic liver disease [17] and fulminant hepatic failure [33] were observed and yet the histopathological adequacy rate was still 100%. Taking these findings into account, future studies could perform subgroup analyses and explore any correlations found between different types of comorbidities at initial diagnosis and how the inclusion or absence of such conditions can influence the yield of adequate tissue samples. This will undoubtedly provide more unbiased evaluation when looking at the rate of adequacy in tissue sampling. However, this meta-analysis still demonstrates consistency with the literature, particularly studies that observe this outcome in adult populations. McCarty et al. reported 97.61% for 179 adult patients [35], which reflects the similarly high rate of adequate tissue samples achieved by performing TJLB in pediatric patients from this meta-analysis. While these results are promising, future studies would benefit greatly with an increased sample size in the pediatric population and if further comparisons are drawn with adults to help elucidate the efficacy of TLJB.

From this meta-analysis, the pooled technical success rate was reported to be 99.1%. In particular, five studies demonstrated a 100% technical success rate [17, 31,32,33,34]. With regards to procedural difficulty, TJLB is generally considered more challenging in pediatric patients when compared to adults, largely due to the difference in size of the patient [32]. However, it is believed that TJLB should not be technically difficult, provided that the operator performing the procedure has adequate experience [36]. Oshrine et al., in particular, had all their procedures performed by consultants trained in pediatric interventions [32], which explains their high rate of technical success. In contrast, other studies did not specifically mention the level of expertise of the operator with regard to pediatric interventions, which could explain lower rates of success [14, 23, 30]. Another parameter that may affect the technical aspect of the procedure is the size of the liver itself. A study by Miraglia et al. was conducted on patients who had left lateral split liver transplant (LT). From their experience, greater technical difficulty was reported in these patients compared to non-LT patients due to the absence of all liver segments except for segments II and III [31]. It is important to note that despite the increased difficulty of the procedure, all eight biopsies reported from their study resulted in adequate tissue samples. Additionally, it is important to consider the outcome of the procedure; notably by observing any changes from the initial diagnosis to the final diagnosis. Our results revealed that samples obtained from TJLB helped confirm the diagnosis in 41.5% of cases, while 5.26% had a new diagnosis established. The first series of TJLB in pediatric patients reported a change in diagnosis in 30% of children [18]. However, only three [14, 17, 30] of the eight included studies reported any changes in diagnosis. Of importance, TJLB played a significant role in diagnosing graft-vs-host disease [17, 22] and graft rejection in transplant patients [23]. While these changes between the initial and final diagnosis are important in assessing the diagnostic power of TJLB, it must be mentioned that more studies need to report these findings for the efficacy to be fully evaluated. Overall, our pooled technical success rate was based on a total of 374 biopsies and while greater sampling sizes in future studies will be of benefit in validating these findings, TJLB has nevertheless proven to demonstrate its safety and efficacy for clinical use in selected pediatric patients.

In terms of technical equipment, the quick-core needle is often favored because of its automated design and tactility for improved manipulation and maneuvering when handling the device. Studies have demonstrated that the quick-core needle yields a high rate of adequate cores and decreases the risk of potential complications in comparison to other types of needles [22, 37]. Five of the included studies in this meta-analysis reported using the quick-core needle for sampling [14, 17, 23, 30, 33]. Habdank et al. report switching from the Colapinto needle to the Quick-core needle after 11 procedures following complications associated with the Colapinto needle [14]. In terms of needle size, Kalambokis et al. stated that thinner needles provided longer and less fragmented specimens [15]. This, in turn, enhances the histopathological quality of the specimens. However, Tran et al. did not report a significant difference between 18 and 19G in terms of fragment numbers and length of samples [33]. Future studies comparing the difference in yield based on the selection of needles and different gauges could provide further insight into this matter and confirm such findings.

Furthermore, 49 complications (13.57%) were reported across all eight studies. Among the complications, the most commonly encountered was minor bleeding from the neck access site (n = 19, 38.78%). In comparison, McCarty et al. recorded only two adverse events following 41 adult TJLB procedures (4.88%) from their meta-analysis [35]. The difference in sample size between the adult population and the pediatric population in this study, as well as the fact that only three of their studies reported adverse events, could have accounted for this discrepancy in complication rate. Miraglia et al. suggested that the wider use of ultrasound guidance has contributed to a reduction in the risk of complications relating to the jugular venous access point. Consequently, no postprocedural complications were reported in their study [31]. Similarly, Habdank et al. reported lower complications after the introduction of sonographic and fluoroscopic guidance compared to before [14]. As such, both findings suggest that imaging has played a major role in association with TJLB in decreasing the rate of complications. Moreover, the literature reports no relation between the number of passes and number of complications [14, 18]. In fact, the meta-analysis by Kalambokis et al. suggests that one of the advantages TJLB holds against the percutaneous approach is that it allows multiple passes with no increase in complication rates [15]. Further studies comparing percutaneous and transjugular liver samplings are necessary to validate this statement. Kalambokis et al. also reports a correlation between smaller liver size and increased risk for perforation. Additionally, this same study demonstrated a 0.6% mortality rate in pediatric patients. In comparison, our results showed a 3.6% (n = 13; 361 patients) mortality rate across studies. However, 12 of the deaths reported in our meta-analysis were not attributed to the procedure [14]. When only considering the procedural mortality rate, this was estimated to be 0.27%. These results suggest that procedural mortality rates may be lower in children, however additional studies comparing adult and pediatric populations are needed to support this.

This meta-analysis offers interesting insight into the usefulness of TJLB in clinical practice. It allows the authors to make the following guidance advice: Firstly, strong evidence is given to the correlation between the use of radiological guidance during TJLB and lowering the rates of complications; Secondly, as the literature reports no correlation between an increase in the number of passes and the rates of complications, practitioners should have no tangible concern to deliberately decrease their number of passes for fear of causing adverse events.

This meta-analysis was not devoid of limitations. Firstly, all the included studies are retrospective, which removes the possibility of randomization. Additionally, most of the studies were conducted in a single-center setting. Studies conducted at multicenter facilities would not only increase the sample size, but they would also add to the reproducibility of the results. The authors made no attempt to include unpublished data from large pediatric institutions, and as such there is possibility of publication bias. Furthermore, no exact indications exist in terms of platelet count and INR values to favor the transjugular route while performing a liver biopsy in pediatric patients [14, 23]. The difference in average platelet counts and INR at baseline could have contributed to heterogeneity in the reported outcomes. Similarly, it is challenging to account for the differences between centers in how they regulate protocols, choice of equipment, intervention technique, and the experience of physicians, therefore also influencing heterogeneity. It should be noted that most of the procedures described in our study employed standard adult TJLB equipment [38]. Finally, our results account for a pediatric population of 14 years and 40 kg on average, thus not accounting for younger patients of lower weight. Despite these limitations, our meta-analysis demonstrates that TJLB in children should be considered as an effective method that can yield high-quality liver samples with little complications at a relatively high technical success rate. As such, future studies would benefit by addressing the above limitations when further evaluating the safety and efficacy of TJLB in pediatric patients.

Conclusion

To our knowledge, this is the first meta-analysis to examine the safety and efficacy of TJLB in pediatric populations. Overall, TJLB has shown to produce adequate quality liver specimens with high rates of technical success and relatively low postprocedural complication rates. While there was heterogeneity among baseline patient characteristics, operator experience and choice of technical equipment, the trends observed in this meta-analysis seem to be in favor of TJLB. In the future, prospective multicenter studies comparing pediatric to adult TJLB are required to help further strengthen the safety and efficacy of TJLB.