Cranial Ultrasound Abnormalities in Small for Gestational Age or Growth-Restricted Infants Born over 32 Weeks Gestation: A Systematic Review and Meta-Analysis

Aim: To perform a systematic review and meta-analysis of existing literature to evaluate the incidence of cranial ultrasound abnormalities (CUAs) amongst moderate to late preterm (MLPT) and term infants, affected by fetal growth restriction (FGR) or those classified as small for gestational age (SGA). Methods: A systematic review methodology was performed, and Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) statement was utilised. Descriptive and observational studies reporting cranial ultrasound outcomes on FGR/SGA MLPT and term infants were included. Primary outcomes reported was incidence of CUAs in MLPT and term infants affected by FGR or SGA, with secondary outcomes including brain structure development and growth, and cerebral artery Dopplers. A random-effects model meta-analysis was performed. Risk of Bias was assessed using the Newcastle-Ottawa scale for case–control and cohort studies, and Joanna Briggs Institute Critical Appraisal Checklist for studies reporting prevalence data. GRADE was used to assess for certainty of evidence. Results: Out of a total of 2085 studies identified through the search, seventeen were deemed to be relevant and included. Nine studies assessed CUAs in MLPT FGR/SGA infants, seven studies assessed CUAs in late preterm and term FGR/SGA infants, and one study assessed CUAs in both MLPT and term FGR/SGA infants. The incidence of CUAs in MLPT, and late preterm to term FGR/SGA infants ranged from 0.4 to 33% and 0 to 70%, respectively. A meta-analysis of 7 studies involving 168,136 infants showed an increased risk of any CUA in FGR infants compared to appropriate for gestational age (AGA) infants (RR 1.96, [95% CI 1.26–3.04], I2 = 68%). The certainty of evidence was very low due to non-randomised studies, methodological limitations, and heterogeneity. Another meta-analysis looking at 4 studies with 167,060 infants showed an increased risk of intraventricular haemorrhage in FGR/SGA infants compared to AGA infants (RR 2.40, [95% CI 2.03–2.84], I2 = 0%). This was also of low certainty. Conclusions: The incidence of CUAs in MLPT and term growth-restricted infants varied widely between studies. Findings from the meta-analyses suggest the risk of CUAs and IVH may indeed be increased in these FGR/SGA infants when compared with infants not affected by FGR, however the evidence is of low to very low certainty. Further specific cohort studies are needed to fully evaluate the benefits and prognostic value of cranial ultrasonography to ascertain the need for, and timing of a cranial ultrasound screening protocol in this infant population, along with follow-up studies to ascertain the significance of CUAs identified.


Background
Cranial ultrasonography is an easily accessible, well-accepted, first-line imaging modality that is widely used to screen preterm and high-risk infants for brain injury Brain Sci. 2022, 12, 1713 3 of 23 affected by fetal growth restriction or classified as SGA. Given FGR is inconsistently defined throughout the literature, Fetal growth restriction (FGR)/Intrauterine growth restriction or SGA (≤10th centile for birthweight) have been used as surrogate markers for growth restriction. Hence, for the purpose of this study, FGR refers to infants classified as either FGR/IUGR or SGA unless specified.

Methods
A systematic review methodology was chosen to investigate what is known about the incidence of abnormal cranial ultrasonography findings amongst infants born from 32 weeks to term gestation and affected by FGR and identify gaps for further research. The (Preferred Reporting Items for Systematic Review and Meta-Analyses) PRISMA statement (Code: CRD42022341804) was utilised [29]. Results were presented as a metaanalysis and synthesis of evidence. The research protocol was registered on PROSPERO (CRD42022341804).

Eligibility Criteria
(1) Inclusion criteria Studies reporting any cranial ultrasonography abnormalities amongst growth-restricted and/or SGA infants published were included. No search filters for publication type or date range were used to ensure a comprehensive search. The results were limited to studies published in English/with English translation and those conducted in humans. Studies were manually reviewed to see if infants from 32 weeks' gestation onwards were included. Case-control, cohort studies and descriptive studies were included. Studies that presented findings for neonates born out of the gestational age criteria but provided subgroup analysis for relevant study population were included. Authors were contacted if subgroup analysis was not available, and if data was provided were included in the review.
(2) Exclusion criteria Studies looking at fetal populations and incidence of antenatal cranial ultrasound findings were excluded, as the purpose of our study was to ascertain if postnatal imaging should be performed. Studies looking at neonates with specific disorders or conditions were excluded (e.g., monochorionic diamniotic twins post laser, infants with congenital heart disease, other congenital anomalies, or metabolic disorders, etc.), as the overall incidence of abnormalities would be higher and not applicable to the general neonatal population. Review articles and case series were excluded, with reference lists of relevant articles crosschecked.

Population
• Moderate-late preterm and term infants, i.e., infants born from 32 weeks' gestation to term age.

Comparator/Control
• If available, to appropriately grown infants of the same gestation group.

Outcome Measures 2.6. Primary Outcomes
Incidence of any cranial ultrasound abnormalities, assessed before discharge as defined by the authors, in MLPT and term infants affected by FGR or SGA, if possible characterised by abnormality: white matter injury, intraventricular haemorrhage, cerebellar haemorrhage, cerebral haemorrhage, structural abnormalities, etc.

1.
Review of brain structure development and growth on cranial ultrasounds, of MLPT and term growth-restricted infants prior to discharge, by assessing: a. 2D-measurements of specific brain structures i. cerebellar vermis size ii.
Review of cerebral artery Dopplers parameters including MCA peak systolic velocity, end diastolic velocity and resistive and pulsatility indices, prior to discharge on cranial ultrasounds of MLPT and term growth-restricted infants All studies reporting cranial ultrasonography findings amongst growth-restricted/SGA infants until December 2021 were included.

Study Selection
All references were imported into Covidence Systematic Review Software (Veritas Health Innovation, Melbourne, Australia) for further review and analysis. Duplicates were automatically excluded by the software. Title and abstract screening and crosschecking of the reference lists of relevant studies was performed by two authors (CR, AM) independently. Full text papers were screened by two authors independently (CR, AM) for subgroup analysis of gestation and weight.

Data Extraction
Data was collected through Covidence with focus on gestation, study type, study numbers, definition of growth restriction, rates of growth-restricted infants, rates of CUAs as total number and percentage for growth-restricted infants and controls if available, types of CUAs identified, key study outcomes and relevant findings to infants either SGA or growth-restricted. Data extraction form is detailed in Appendix SB.

Analysis
The papers identified through consensus were included in qualitative and quantitative analysis. Result findings were reported using the PRISMA statement for systematic reviews.
A random-effects model meta-analyses was performed using Review Manager 5.4.1 (Cochrane Collaboration, Nordic Cochrane Centre, Copenhagen, Denmark) to yield pooled odds ratio (ORs) and 95% Confidence Incidence (CI) for any CUAs and IVH specifically. A p-value of <0.05 was considered as significant. Studies comparing MLPT and late preterm to term were pooled as subgroups to test for subgroup differences and identify the source for heterogeneity. Statistical heterogeneity was reported using I 2 values (derived from the χ 2 Q-statistic). Significant statistical heterogeneity was an I 2 value greater than 50% or p for χ 2 < 0.10.
The Newcastle-Ottawa scale [30] was utilised to assess the quality of studies and the risk of bias by looking at domains of selection, comparability, and outcomes. Refer to Appendix SC. For descriptive studies the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for studies reporting prevalence data [31] was utilised.
Publication bias assessment was not possible as less than 10 studies were included in the meta-analysis. Assessment of the certainty of evidence was performed using the Cochrane Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) methodology.

Search Results
A literature search was conducted using search terms as described above. Figure 1 depicts the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) flow diagram for the systematic review.
The search returned 2027 citations, 329 from MEDLINE, 877 from EMBASE, 763 from PubMed and a further 58 through manual reference checking of review articles. 568 duplicates were removed, 565 were removed through Covidence Systematic Review Software (Veritas Health Innovation, Melbourne, Australia) and another 3 were removed manually. Abstracts from 180 studies were assessed for eligibility, which led to the inclusion of 17 studies that contained relevant data for infants of interest, 14 studies looked at CUAs, 3 studies looked at brain growth and Dopplers. Of the 14 studies looking at CUAs, a further 7 studies looked at FGR verses AGA infants and lent themselves to meta-analysis.

Methodological Quality
The Newcastle-Ottawa scale was used to assess the risk of bias in case-control and cohort studies. [30] Nine studies had a moderate risk of bias [32][33][34][35][36][37][38][39][40], four studies had high risk of bias [41][42][43][44], and one study had a low risk of bias [45]. (Refer to Tables 1 and 2 for Newcastle-Ottawa scale scoring). The studies that had issues with high risk of bias had issues in all three domains: selection, comparability, and outcomes. Issues around selection criteria, with many studies not mentioning the number of eligible subjects, may have led to response bias. The main issue with comparability was not controlling for weight or other confounding factors. Issues with outcomes identified were that the studies had variable timing of when cranial ultrasounds were preformed, with some only reporting very early ultrasounds leading to possible reporting bias and missing abnormalities that develop or persist. Publication bias assessment was not possible.
The certainty of the evidence as assessed per GRADE was deemed very low due to the risk of bias, heterogeneity, and observational data.

Study Characteristics
The characteristics of the studies included were six prospective cohort studies and seven retrospective cohort studies and one case-control study. The main findings of articles reviewing MLPT infants are listed in Tables 3-5     One star was awarded for each domain, except for comparability, where a maximum of two stars could be awarded. Total scores ranged from 0-9. Low risk of bias studies were awarded 7-9 stars, moderate risk of bias studies were awarded 4-6 stars and high risk of bias studies were awarded 0-3 stars. One star was awarded for each domain, except for comparability, where a maximum of two stars could be awarded. Total scores ranged from 0-9. Low risk of bias studies were awarded 7-9 stars, moderate risk of bias studies were awarded 4-6 stars and high risk of bias studies were awarded 0-3 stars. To determine morbidity and mortality in growth-restricted infants with early onset placental dysfunction Gestational age was the most significant predictor (p < 0.005) of survival until 26 +6 weeks and intact survival until 29 +2 weeks.
Beyond 29 +2 weeks and in infants > 600 g, ductus venosus Doppler and cord artery pH was predictive of neonatal mortality (p < 0.001) and Doppler alone was predictive of intact survival.
Of 76 infants born growth-restricted at 32 weeks 1.3% (n = 1) had a grade III-IV IVH        Of the seven studies included in the meta-analysis, three were prospective cohort studies, three were retrospective cohort studies and one was a case-control study.

Cranial Ultrasound Abnormalities in Moderate to Late Preterm Infants
Ten studies assessed CUAs in MLPT growth-restricted infants (see Tables 3 and 4) [32,34,38,40]. Six studies compared growth-restricted to well grown infants [32,34,37,38,40,45], with four of these studies showing an increased risk of abnormalities in growth-restricted infants [32,34,38,40]. Of these studies seven studies reported only IVH changes on cranial ultrasound with one of these studies only reporting Grade III/IV IVH [46] (see Table 5 for further details).
When looking at which individual CUAs were reported, four studies reported IVH as the only CUA [32,34,38,40]. The meta-analysis of these 4 studies included 167,060 participants and showed an increased risk of IVH in FGR infants compared to AGA infants (RR 2.40, [95% CI 2.03-2.84], I 2 = 0%). (Refer to Figure 4) There was no heterogeneity. The evidence certainty was low due to non-randomised studies and their methodological limitations.
was performed looking at MLPT (32-37 +6 weeks) and late preterm to term (36-42 weeks) infants, respectively. An increased risk of CUAs was found in the late preterm to term subgroup (RR 4.34, [95% CI 1.18-15.95]) but not in the moderate to late preterm subgroup (RR 1.96, [95% CI 0.96-4.13]) and the test for subgroup differences were not significantly different (p = 0.31) (Figure 3). When looking at which individual CUAs were reported, four studies reported IVH as the only CUA [32,34,38,40]. The meta-analysis of these 4 studies included 167,060 participants and showed an increased risk of IVH in FGR infants compared to AGA infants (RR 2.40, [95% CI 2.03-2.84], I 2 = 0%). (Refer to Figure 4) There was no heterogeneity. The evidence certainty was low due to non-randomised studies and their methodological limitations.  No separate data was available to perform a meta-analysis of other CUAs, such as PVL, cerebral or cerebellar haemorrhage.

Brain Structure Growth in SGA Infants
There has been some interest in cerebellar size in infants with growth restriction as a potential prognostic indicator. Huang looked at whether there was a difference in the growth of the cerebellar vermis between AGA and SGA infants [47]. There was a significant difference in the central vermian area and the superior-inferior distance of the cerebellar vermis in infants born 38-41 weeks. Makhoul looked at transverse cerebellar diameter in preterm and term infants that were AGA verses SGA [48]. They found that in both asymmetric and symmetric SGA infants there was not a statistically different size in transverse cerebellar diameter in SGA verses AGA infants.

MCA Dopplers in Infants with Fetal Growth Restriction
Krishnamurthy et al. looked at MCA Doppler characteristics in 40 infants comparing 20 with FGR and 20 AGA infants [49]. They performed a subgroup analysis and infants >32 weeks gestation and found a higher MCA peak systolic velocity in FGR infants on day 1 (p = 0.01) and 3 (p = 0.009) of life. The end diastolic velocity value was only significant on day 3 (p = 0.007) of life in FGR infants. In infants > 32 weeks the resistive index and pulsatility index were significantly lower in FGR compared to AGA infants.

Discussion
This systematic review and meta-analysis provides an overview of the literature on CUAs in MLPT and term growth-restricted/SGA infants, and to our knowledge is the first of its kind. When reviewing the evidence, our aim was to ascertain what is known No separate data was available to perform a meta-analysis of other CUAs, such as PVL, cerebral or cerebellar haemorrhage.

Brain Structure Growth in SGA Infants
There has been some interest in cerebellar size in infants with growth restriction as a potential prognostic indicator. Huang looked at whether there was a difference in the growth of the cerebellar vermis between AGA and SGA infants [47]. There was a significant difference in the central vermian area and the superior-inferior distance of the cerebellar vermis in infants born 38-41 weeks. Makhoul looked at transverse cerebellar diameter in preterm and term infants that were AGA verses SGA [48]. They found that in both asymmetric and symmetric SGA infants there was not a statistically different size in transverse cerebellar diameter in SGA verses AGA infants.

MCA Dopplers in Infants with Fetal Growth Restriction
Krishnamurthy et al. looked at MCA Doppler characteristics in 40 infants comparing 20 with FGR and 20 AGA infants [49]. They performed a subgroup analysis and infants > 32 weeks gestation and found a higher MCA peak systolic velocity in FGR infants on day 1 (p = 0.01) and 3 (p = 0.009) of life. The end diastolic velocity value was only significant on day 3 (p = 0.007) of life in FGR infants. In infants > 32 weeks the resistive index and pulsatility index were significantly lower in FGR compared to AGA infants.

Discussion
This systematic review and meta-analysis provides an overview of the literature on CUAs in MLPT and term growth-restricted/SGA infants, and to our knowledge is the first of its kind. When reviewing the evidence, our aim was to ascertain what is known about the incidence of CUAs in MLPT and term FGR/SGA/IUGR infants and if routine screening should be considered. It is apparent, that there are still many gaps in the literature. Four main findings of the meta-analysis were that when the data was pooled, infants MLPT and term (>32 weeks) were found to have an increased rate of CUAs in FGR infants when compared with AGA infants. When further looking at subgroup analysis, moderate-late preterm infants born FGR did not have a significant increased rate of CUAs compared with their AGA counterparts; however, late preterm, and term growth-restricted infants had an increased rate of CUAs. When looking at IVH alone, there was an increased rate in infants born FGR versus AGA. Overall, however, the studies presented were heterogenous, of varying quality, and the certainty of evidence was overall low.
Overall, there was a paucity of studies focusing on CUAs in MLPT and term infants, especially looking at FGR as a risk factor. Many of the studies selected were not set up to answer the clinical question and were often subgroup analyses with small cohorts potentially misrepresenting the true incidence of CUAs in this population group. The definition of FGR varied widely from ICD coding, varying birth weight cut offs (e.g., <3rd, <5th and <10th centiles) and weight and Doppler abnormalities. Even more difficult to interpret is the wide variety in incidence of cranial ultrasound abnormality results, at times greater than those of infants in the extreme and very premature category. While ultrasound is known to be operator-dependent, further challenges around consensus between radiologists and neonatologists when reporting and defining ultrasound findings all add additional complexities when interpreting the data.

Incidence of Cranial Ultrasound Abnormalities
The incidence of CUAs varied significantly when reviewing the literature. Marsoosi et al. reported the highest rate (33%) of IVH; however, they looked at 6 infants born between 32 to 33 +6 weeks and found that 2 of these infants had IVH. Such a small sample size brings into question the validity of such a high incidence of abnormalities. Ballardini reported the second highest incidence of CUAs with 19.4% in <3rd centile infants born between 33 +0 -36 +6 weeks; PVL was not defined, and increased incidence could be due to potential over-reporting of white matter injury [45,50]. Gilbert reported the lowest incidence of IVH in FGR MLPT infants, with an incidence of 0.4% in infants born at 36 weeks [34]. While this was a comprehensive and extensive database review, the difficulty with interpreting these findings is that FGR was defined by ICD coding. Baschat reported the second lowest incidence in this cohort of 1.3% in infants born 32-32 +6 weeks, however, they only reported grade III-IV IVH [41]. Understandably these are the most significant changes on ultrasound linked with long-term neurodevelopmental complications; however, more recent evidence is suggesting that even grade I and II IVH may indeed have long-term complications due to microstructural changes in periventricular and subcortical white matter [51].
Similarly in the studies looking at late preterm and term infants the incidence of abnormalities varied widely. Mercuri studied a very small cohort of growth-restricted infants that found 7 of the 10 infants with CUAs, the specific findings of which were not listed and is therefore difficult to interpret. Surprisingly the most common abnormality documented was ischaemic lesions with 8% infants having periventricular and thalamic densities. Starcevic's rate of CUAs of 53.37% is also hard to explain. Infants were screened on day 1, 3 and 7 and while PVL changes may have been overestimated, the rate of IVH was also significantly higher than recorded rates of very preterm infants and whether differences around perinatal care may affect these rates so drastically is questionable [51].
Our findings of such variability in incidence reported in the literature is not unique, a recent systematic review by Boswinkel looked at brain injury and altered brain development in moderate-late preterm infants, by reviewing a mix of cranial ultrasound and MRI results and found a wide incidence of CUAs (0.0 to 23.5%) [50]. There was noted to be paucity of high-quality studies set up to answer the incidence of brain injury in this population and heterogenous study cohorts.

FGR versus AGA Cranial Ultrasound Result Comparison
When comparing FGR infants with AGA infants MLPT infants did not have a significant risk of CUAs, whereas late preterm and term FGR infants may have an increased risk of CUAs. (See Figure 3). When looking at MLPT infants, four of six studies identified showed an increased risk of CUAs in MLPT, three [32,38,40] used SGA as a surrogate for FGR and one used ICD coding [34]. When looking at late preterm and term infants, two of the three studies identified showed an increased rate of CUAs. One study used ICD coding, a further used a cut-off of weight < 10th centile and yet another referred to medical notes.
These results may suggest that infants born MLPT may indeed have higher rates of CUAs than once thought, additionally using SGA as a surrogate marker may not accurately capture those infants affected by FGR and render any difference between cohorts as insignificant. Of the two studies that found FGR infants were not at increased risk of CUAs, Ballardini and colleagues overall found a high rate of CUAs in their cohort and concluded that screening should be considered for the MLPT population, especially the 33-34-week cohort [45]. While Medina-Alva et al.'s purpose was not to look at the incidence of CUA in our intended population, and indeed the rate of major abnormalities on CUA was slightly greater in those infants born AGA compared to SGA [37]. This discrepancy may be explained by the perinatal complications surrounding infants born MLPT where there may be chorioamnionitis or incomplete steroid use, compared with the planned and controlled delivery of SGA infants, alternatively it may be due to the use of SGA as a cut off and not truly identifying FGR infants.
When looking at late preterm and term growth-restricted infants, the two studies that showed an increased rate of CUA in growth-restricted infants were significant outliers with their reported incidence of abnormalities [33]. Conversely, Berger found that SGA infants at term were not at increased risk of CUA, this may be explained by a heterogenous cohort of healthy small and growth-restricted infants [33].
While the focus has traditionally been on IVH and white matter injury; our understanding of the CUA seen in MLPT and term infants, especially those that are growth-restricted, is limited. Our focus may need to shift from our typical patterns of brain injury seen in prematurity, to looking at brain growth of specific structures, Doppler parameters, or consideration of MRI to further delineate more subtle patterns of brain injury.
Overall, MLPT and term growth-restricted infants are potentially no longer the lowrisk cohort we once believed. Various other risk factors have been noted to contribute towards intracranial pathology in the late preterm population-including intubation, low cerebroumbilical ratios, hypothermia, low APGAR scores, head circumference below the 3rd percentile at birth, low venous/arterial pO2 and pH [32,36,45,52]. Given the association between increased rates of intracranial pathology and low arterial/venous pH and pO2, and use of ventilatory supports, there may be suggestion that hypoxic insult in MLPT infants plays an important indicator contributing towards abnormal cranial ultrasonography results. Starcevic et al. also described low cerebroumblical ratios (CUR), umbilical resistance index and low birthweight to be related to adverse functional outcomes indicating clinical significance. CURs ≤ 1 are an important marker of fetal compromise and have been linked to adverse perinatal outcomes with strongest association in low birthweight babies [53]. Khazardoost also found that absent or reversed end-diastolic velocity in the UA and DV was associated with an increased rate of CUAs [54]. Although it is not possible to fully ascertain the benefit of cranial ultrasonography screening upon all growth-restricted infants, risk stratification may be a useful tool in the development of screening guidelines. The presence of small-for-gestational age, low cerebroumbilical ratios and indicators of birth asphyxia may act as important tools to identify high risk infants that may benefit from cranial ultrasound screening.

Limitations
To date, there have been limited studies focusing on MLPT and term growth-restricted infants and minimal data evaluating long-term developmental outcomes associated with abnormal cranial ultrasonography results. Inconsistent definitions of growth restriction and the use of surrogate markers to identify growth restriction present a challenge in recognition of the pathologically growth-restricted fetus and may attribute to the discrepancies seen in the literature. Furthermore, this review carries multiple limitations including significant heterogeneity of studies, methodologically weak studies not powered to answer our research question and low certainty of the evidence. When addressing heterogeneity, multiple aspects may have contributed to this including differences in risks and screening, pooling of CUAs data together and gestation subgroups. There was no heterogeneity when looking at IVH alone, suggesting that pooling all CUAs abnormalities may be one possible source of heterogeneity. There were also considerable differences in gestational age among studies and despite no difference between the subgroups, merely testing two subgroups may not be sufficient. Additionally, studies were observational in nature, retrospective and with differing study aims, with data often being extrapolated to examine specific subgroups of interest. Studies were of varying designs, with different cohort years and protocols for and timing of cranial sonography. Ultrasound screening protocols varied between studies, and the timing and frequency may play an important role in the detection of intracranial abnormalities. Eight studies alone focused on IVH and no other CUAs, which may not in fact be the focus when imaging MLPT and term infants, conversely pooling all CUAs lead to significant heterogeneity. In addition, cohorts were grouped in moderate-late preterm and late preterm-term infants as many studies did not gave further subgroup data available or had overall small cohorts. This further complicates the interpretation of data infants as each gestational group is a unique entity with often different reasons for delivery and different risks of brain injury.

Implications for Future Research
Dedicated cohort studies with multivariable analysis are needed to fully evaluate the benefits and prognostic value of cranial ultrasonography to ascertain the need for and timing of a cranial ultrasound screening protocol in this infant population. Further research on brain growth and MCA Dopplers in further characterising structural changes associated with FGR that may be useful for prognostication. Long-term follow-up studies to ascertain the significance of CUAs in the MLPT and term growth-restricted cohort are warranted. The development and evaluation of risk stratification tools may play an important role in developing neuroimaging guidelines for identification of infants that are at greater risk of neurodevelopmental abnormalities.

Conclusions
The incidence of CUAs in MLPT and term growth-restricted infants varied widely between studies. The findings from the meta-analysis suggest the risk of CUAs and IVH may indeed be increased in FGR/SGA infants when compared with infants not affected by FGR, however the evidence is of low to very low certainty. It is increasingly apparent that these cohorts have long-term neurological sequelae based on long term follow up studies; however, the incidence, pattern of brain injury and potential abnormalities on neuroimaging in addition to their long-term outcomes have been poorly characterised thus far.