|Home | About | Journals | Submit | Contact Us | Français|
To identify risk factors for cerebral lesions among survivors of TTTS treated with laser surgery.
A multilevel regression analysis examined risk factors for neonatal cerebral lesions identified by imaging. Imaging was routine in “high-risk survivors”, defined as those delivered at <32 weeks' gestation, and by clinical indications if born later. Severe lesions were defined as: intraventricular hemorrhage grade III–IV, cystic periventricular leukomalacia, ventriculomegaly and/or hydrocephalus, microcephaly, infarctions, porencephalic/Dandy-Walker cysts, or bilateral other cysts.
For 262 consecutive laser-treated TTTS patients, 18 neonates had severe lesions identified among 427 individual survivors (4.2%) and 242 “high-risk survivors” (7.4%). Forty-six newborns had any cerebral lesion, resulting in lesion rates of 10.8%–19.0%. Delivery <32 weeks' (OR=4.95, p<0.001) and <28 weeks' (OR=6.25, p<0.001) gestation were associated with increased likelihood of any cerebral lesion.
This cohort showed low rates (4–7%) of severe neonatal cerebral lesions, with prematurity being the primary risk factor.
Twin-twin transfusion syndrome (TTTS) occurs in monochorionic twin gestations due to unbalanced exchange of blood through placental vascular communications. The optimal treatment for TTTS is selective laser photocoagulation of communicating vessels (SLPCV), which has led to improved single and dual twin survival.1–3 With this improvement in survival, concern exists regarding risk factors to neonatal neurological status and later neurodevelopmental outcomes.
There is a relative paucity of data regarding neurological outcomes after laser surgery, particularly involving U.S. patients. In a systematic review and meta-analysis of neurodevelopmental outcomes after laser therapy for TTTS published in 2011, seven articles were identified that described neonatal neurological outcomes.4 Overall, the prevalence of neurologic morbidity, defined as cerebral injury on imaging, cerebral palsy, blindness, and/or deafness, was 6.1%.4 Studies with particular emphasis on neonatal cerebral imaging have shown rates of severe cerebral lesions of 5–14%.5–7 The 2011 meta-analysis also showed that the prevalence of long-term neurodevelopmental impairment was 11.1%,4 with rates of cerebral palsy in the range of 4–6%.8–9 Thus, counseling of prospective TTTS patients contemplating laser surgery must take into account survival statistics as well as potential neurological morbidity. Elucidation of the types of cerebral lesions and the antecedent factors would enhance parental counseling and decisions regarding treatment.
The aim of this study was to determine the prevalence of neonatal cerebral lesions and risk factors among TTTS survivors treated with laser surgery in a large cohort of U.S. patients.
All consecutive patients that were treated for TTTS between March 2006 and May 2011 were considered eligible for this retrospective study. TTTS was diagnosed at initial assessment at Los Angeles Fetal Therapy (University of Southern California) if the monochorionic-diamniotic multiple gestation had a maximum vertical pocket of fluid ≥ 8 cm in the recipient's sac and ≤ 2 cm in the donor's sac. Each case was prospectively classified according to the Quintero staging system.10 Patients were not offered laser surgery if preoperative ultrasound revealed abnormal intracranial findings, including intraventricular hemorrhage, porencephalic cysts, hydrocephalus, isolated ventriculomegaly (atrium >20 mm), or congenital anomaly including Dandy-Walker syndrome, holoprosencephaly, and agenesis of the corpus callosum. All cases were exclusively treated via selective laser photocoagulation of communicating vessels (SLPCV) with or without sequential technique, as described in detail previously.11 Umbilical cord occlusion is not a treatment option for TTTS at this center.
After laser surgery, the patients were referred back to their perinatologists for further pregnancy management. Delivery and 30-day survival information were collected at their respective intervals. Placental pathological evaluation and injection studies were performed using previously reported methodology.12 Medical records of delivery, neonatal course, and all identified imaging reports were retrieved and reviewed. With few exceptions, neonates with a gestational age (GA) < 32 weeks had routine cerebral imaging, and neonates with a GA ≥ 32 weeks had cerebral imaging when clinically indicated or deemed appropriate by the NICU team. “High-risk survivors” were defined as 1) those delivered at a GA <32 weeks, and 2) those born later for whom cerebral imaging was performed because of a clinical indication. Cerebral imaging was defined as a post-natal intracranial ultrasound, CT scan, and/or MRI. Reported results defaulted to the confirmatory scans. This study was approved by the Institutional Review Board of the Health Sciences Campus of the University of Southern California.
Potential risk factors measured at the individual (i.e., fetal) level included donor status (1=donor, 0=recipient) and birth weight (kg). Potential risk factors measured at the pregnancy level (i.e., measurements that would be identical within a pair of twin fetuses) included Quintero stage (ordinal with range 1 to 4); gestational ages at diagnosis, surgery, and birth (in weeks); existence of a triplet (1=triplet pregnancy, 0=no triplet); co-twin intrauterine fetal demise (1=co-twin fetal demise, 0=no fetal demise); and presence of residual vascular anastomoses via placental injection studies (1=residual anastomoses, 0=no residual anastomoses). Gestational age at birth was also recoded into a binary indicator of prematurity under 32 weeks (1=age < 32 weeks, 0=age ≥ 32 weeks) and under 28 weeks (1=age < 28 weeks, 0=age ≥ 28 weeks). All “singletons” of triplet gestations (i.e. the triplet not involved in the feto-fetal transfusion) were excluded from the analysis; however, the analysis did include twin survivors of triplet gestations.
For the cerebral lesion outcomes, “any lesions” were defined as: intraventricular hemorrhage (IVH), cystic periventricular leukomalacia (PVL), ventriculomegaly and/or hydrocephalus, microcephaly, single or multiple infarctions, congenital anomalies, porencephalic or Dandy-Walker cysts, nonspecific echogenicity, and bilateral/multiple subependymal, pseudo-, or choroid plexus cysts identified on neonatal imaging (coded 1=any lesion, 0=no lesion). “Severe lesions” excluded cases with only grade I–II IVH and/or nonspecific echogenicity.13–15 Unilateral subependymal, pseudo-, and choroid plexus cysts were not included in “any lesions” and excluded from the analysis as incidental findings.16
SPSS (version 19) was used for data management and reporting of univariate statistics. Mplus software (version 6) was used to perform multilevel logistic regression analysis,17 also known as hierarchical linear modeling or hierarchical regression analysis,18–21 with twins grouped or “nested” within pregnancy. In the present study, a multilevel approach accounts for shared variance within twin pairs, which is created by pregnancy and family factors that are equivalent among the twins, as well as for heterogeneity between pregnancies and families. In this way, multilevel analysis allows us to simultaneously model the effects of both individual (i.e., fetus-level) and pregnancy-level risk factors on the likelihood of cerebral lesion development.
Regression models initially included donor status and birth weight at the individual level and Quintero stage, gestational ages at diagnosis, surgery, and birth, existence of a triplet, co-twin demise, and presence of residual vascular anastomoses at the pregnancy level. Due to the very low incidence of severe cerebral lesions among neonatal survivors in this cohort, risk factors to severe lesion development were not investigated further. Thus, only the presence of any lesion was regressed on the individual- and pregnancy-level risk factors. Risk factors to any lesion development were first tested individually (unadjusted model) and then simultaneously (adjusted model), and non-significant predictors were systematically removed from the adjusted model.
A total of 262 consecutive TTTS cases were treated via SLPCV between March 2006 and May 2011. Of the 262 twin gestations, 242 (92%) had at least one neonatal survivor and 185 (71%) had dual survivors at 30 days. All demises prior to 30 days after birth, whether intrauterine (N=78) or in the neonatal period (N=19), were excluded from the analysis. Of the 427 “total survivors”, 137 were delivered at a gestational age of less than 32 weeks, and 134 of those had routine cranial ultrasounds. Of the remaining 290 30-day survivors that were born at or beyond 32 weeks' gestation, 108 had cerebral imaging due to other clinical indications, and 182 had no clinical indication for imaging and thus did not undergo an imaging procedure. Thus, among our sample, 242 “high-risk survivors” had delivery, demographic, 30-day survivorship, and imaging information available for analysis (see Figure 1).
Among the entire cohort of 427 survivors, 46 (10.8%) had a documented cerebral lesion and 18 (4.2%) had a documented severe lesion if survivors with only grade I–II IVH and/or nonspecific echogenicity were excluded. All survivors with a documented cerebral lesion (N=46) had a cranial ultrasound except for one who had magnetic resonance imaging only. Of the 45 survivors with a cranial ultrasound, 12 had an additional follow-up magnetic resonance imaging and 2 others had follow-up computed tomography scan. Among the 242 “high-risk survivors” indicated for cerebral imaging, the rates for any cerebral lesion and severe cerebral lesion were 19.0% and 7.4%, respectively. In this sub-group of “high-risk survivors”, the prevalence of any cerebral lesion and severe cerebral lesion in the survivors born between 24 to <28 weeks was 44.7% and 18.4%; between 28 and <32 weeks was 20.8% and 7.3%; and 32 weeks or greater was 8.3% and 3.7%, respectively (see Figure 2).
Table 1 presents descriptive statistics for the risk factors and outcomes by donor/recipient status and overall for the “high-risk survivors”. Multiple lesions were found in several of these survivors. Documented severe lesions included: grade III or IV IVH (N=6; 2.5%), PVL (N=7; 3%), ventriculomegaly/hydrocephalus (N=6; 2.5%), microcephaly (N=1; 0%), Dandy-Walker cyst (N=1; 0%), and bilateral or multiple choroid plexus cysts (N=1; 0%). No survivors were diagnosed with single or multiple infarctions. There were seven survivors with residual anastomoses; among these, two had grade I–II IVH and one had a co-twin with an intrauterine fetal demise (IUFD) but no lesions. None of the survivors with residual anastomoses had severe lesions. No significant difference in the rate of any cerebral lesion were noted in the sub-group of “high-risk survivors” with an IUFD (7.7%) compared to those with dual survivors (20.4%, p=0.18). Similarly, no difference in prevalence of severe cerebral lesions were detected in these two sub-groups (3.8% vs. 7.9%, p=0.70, respectively).
Using the multilevel modeling approach, it is possible to compute an intraclass correlation that represents the proportion of total variation in the outcome present at each level of the model. The intraclass correlation of the empty model for any lesion (i.e., a multilevel model with no predictors) was 0.259, indicating that roughly a quarter of the variation in any lesion development outcome was attributable to pregnancy-level, rather than individual-level, effects, demonstrating the need to account for both sources of variation in modeling survivor outcomes. A series of multilevel logistic regression models were then conducted in Mplus to investigate the relationships, if any, between individual- and pregnancy-level risk factors and likelihood of any lesion. Table 2 presents the unadjusted regression model results from a series of analyses in which each risk factor for any lesion was tested in the model one-at-a-time. Higher birth weight (kg) (OR=0.12, 95% CI=[0.04,0.34], p<0.001) and later gestational age at birth (OR=0.73, 95% CI=[0.63,0.85], p<0.001) were protective. The binary indicator of prematurity defined as birth <32 weeks (OR=4.95, 95% CI=[1.91,12.81], p=0.001) and < 28 weeks (OR=6.25, 95% CI=[2.16–18.07], p=0.001) were significant risk factors for development of any lesion in the “high-risk survivors”.
Subsequently, the effects of all potential individual- and pregnancy-level risk factors were modeled simultaneously in a multivariate multilevel regression model. Separate models were run including either gestational age at birth or the prematurity indicator. In the multivariate model, only birth weight (OR=0.11, 95% CI=[0.04,0.31], p<0.001) remained a significant risk factor for any lesion development. Curiously, triplet pregnancy (OR=0.20, 95% CI=[0.04,0.98], p<0.05) became significantly associated with a lower odds of any lesion after controlling for the variance associated with low birth weight. After accounting for the variance explained by these predictors in the adjusted model, 18.4% of the variation in the occurrence of any lesion remained at the pregnancy rather than individual level.
This US cohort of TTTS twins treated with laser surgery showed rates of severe neonatal cerebral lesions between 4% (entire cohort) and 7% (“high-risk survivors”). Preterm birth and low birth weight were the primary risk factors associated with the detection of any cerebral lesion. This study supports the contention of previous reports that have suggested that neurological morbidity after laser surgery for TTTS results primarily from prematurity related complications.22 These findings suggest that optimization of length of pregnancy may be a key factor to avoiding cerebral lesions in TTTS survivors.
While not deterministic, studies have shown an increased risk of abnormal neurodevelopmental outcomes with the presence of severe cerebral lesions. A meta-analysis among preterm infants found that after a specific abnormality is identified on cranial ultrasound, there was a higher post test probability for the presence of cerebral palsy or neuromotor impairment for grade III IVH (26%), grade IV hemorrhage (53%), cystic PVL (74%), ventricular dilation (22%), and hydrocephalus (27%).23 Despite these increased risks, multiple factors influence neurodevelopmental outcomes, making cerebral imaging part of an overall risk profile.24 Formal neurodevelopmental testing of the individuals in this cohort at two years of age is ongoing.
There were several pertinent negative results in this study. The intrauterine demise of one twin did not put the surviving co-twin at an increased risk of a cerebral lesion, as would be the case in a monochorionic twin gestation that did not undergo laser surgery.25 This would lead to the speculation that SLPCV may be protective in the surviving co-twin, but this study lacked the power and appropriate study design to test this hypothesis. Assuming all vascular communications have been successfully occluded, the laser surgery transforms the pregnancy into a functional dichorionic twin gestation, thereby rendering the circulatory systems of the twins independent of one another. The rate of residual anastomoses was relatively low in this cohort (3%). However, there were seven survivors that had a residual anastomosis after laser surgery; fortunately, none had a severe cerebral lesion, though two showed less significant grade I–II IVH. Second, the lack of effect of Quintero stage was surprising based on previous long-term studies.4 Finally, this study did not find donor/recipient status, or gestational age at diagnosis or surgery, to be significant risk factors.
The primary methodological strength of this study was the relatively large sample size available for analysis. The extensive prospectively collected data of prenatal and postnatal risks were used to determine clinically relevant factors to these immediate neurological outcomes. However, there are several important weaknesses of this study. First, cerebral lesions were not common, thus this study may overemphasize the significance of risk factors common to the few participants who suffered lesions. A larger cohort may indeed identify further risk factors for cerebral lesions (other than prematurity and birth weight) that may be clinically important. Second, the study used clinical data from the various birth hospitals, which could have led to misclassification or omission of lesions based on the expertise of the radiologist. Third, scans were not available on the 182 late preterm or term survivors who had no clinical indication for imaging. Thus, the regression models for any cerebral lesions were performed in the “high-risk survivors”, which may underestimate the final odds ratio for prematurity in this model. On the other hand, there is the possibility that cerebral lesions were not detected in this sub-group. These could lead to over- or underestimation of the prevalence of lesions. We tried to take this into account by reporting a range of severe cerebral lesion rates of 4%, which includes the entire cohort, and 7%, which includes only the “high risk survivors” with clinically indicated imaging. Fourth, data were not available to document the timing of these lesions. Fifth, limitations in the study methodology cannot take into account survivorship bias, or those individuals who may have died with cerebral lesions. Finally, unmeasured covariates could have led to confounding and played a role in creating worse outcomes for some participants.26
In summary, we found a relatively low prevalence (4 to 7%) of severe cerebral lesions in neonatal survivors of TTTS after laser surgery. The primary risk factor for these lesions appears to be prematurity and low birth weight. The clinical relevance of these lesions must be seen in the correlation with later neurodevelopmental outcomes. Further investigation of the presence of cerebral lesions as a risk factor to later neurodevelopmental outcomes is ongoing.
Financial Support: None.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclosure: None of the authors have a conflict of interest.
Presented at the 32nd Annual Meeting of the Society for Maternal-Fetal Medicine, Dallas, TX, Feb. 6–11, 2012.