A novel approach to calculating expected total fetal lung volume in fetuses with isolated congenital diaphragmatic hernia and fetal growth restriction: a theoretical computational simulation

Introduction

Congenital diaphragmatic hernia (CDH) is a complex congenital anomaly that poses significant challenges in neonatal care. CDH results in impaired lung development with resulting pulmonary hypoplasia, making it a life-threatening condition for affected infants [1], 2]. Additionally, fetal growth restriction (FGR) – defined as an estimated fetal weight (EFW) less than the 10th percentile for gestational age (GA) – has been reported in association with CDH in varying proportions, although the exact incidence in isolated CDH cases without chromosomal abnormalities is not well established [3], 4]. Currently, neonates with CDH and FGR are managed similarly to those without FGR, and FGR alone does not preclude eligibility for fetal interventions.

During the prenatal period, several imaging biometric markers have been used for antenatal risk stratification of CDH and for determining eligibility for fetal interventions. Use of magnetic resonance (MR) imaging has been shown to predict outcomes more accurately than ultrasound techniques and also allows for the detection of other associated anomalies [5], [6], [7]. A combination of MRI O/E TFLV and percent liver herniation (%LH) has the best reported accuracy in predicting mortality and the need for Extracorporeal Membrane Oxygenation (ECMO) [7]. The O/E TFLV is classically reported as a percentage that compares the observed TFLV to what is expected in a healthy fetus of the same GA based on one of two widely employed normative formulas, namely the Meyers’ and Rypens’ formulas [8], 9]. However, the accuracy of the O/E TFLV may be compromised in cases of FGR due to its reliance solely on GA without consideration for fetal size.

In this computer simulation study, we examine the potential influence of incorporating a weight-based approach to calculating the expected TFLV (eTFLV). Our approach involves using an adjusted GA based on the EFW and adjusting the normative values of the eTFLV to calculate a new O/E TFLV. Through rigorous analyses and simulations, we sought to evaluate the degree by which this novel approach would change prognostic predictions and eligibility for fetal interventions. Real patient studies, however, are needed to validate this theoretical approach.

Materials and methods

Results

Simulated CDH O/E TFLV calculation using FGR correction

A total of 1,005 simulated patients were generated, which included 67 simulated patients for each GA between 22- and 36-weeks’ gestation. The mean generated OFW, standard O/E TFLV, and %LH along with the mean adjusted GA and O/E TFLV for each GA are listed in Table 1.

Table 1:

Comparison of standard and corrected GA and O/E TFLV for each GA.

GA	EFW (g) based on GA a	eTFLV (mL) based on GA b	Mean OFW, g c	Mean standard O/E TFLV c	Mean %LH c	Mean corrected GA, weeks	Difference in GA, weeks	Mean corrected O/E TFLV	% Difference in O/E TFLV
22	493	20.4	408	0.31	22	21.0	1	0.37	6 %
23	572	24.1	467	0.36	22	21.84	1.16	0.44	8 %
24	662	27.8	541	0.38	21	22.72	1.28	0.46	8 %
25	767	31.6	621	0.39	21	23.57	1.43	0.48	9 %
26	890	34.8	725	0.37	21	24.72	1.28	0.42	5 %
27	1,031	40.3	834	0.37	21	25.63	1.37	0.44	7 %
28	1,192	48.8	978	0.38	22	26.7	1.3	0.48	10 %
29	1,369	54.2	1,117	0.37	21	27.54	1.46	0.44	7 %
30	1,561	63.3	1,284	0.34	23	28.6	1.4	0.42	8 %
31	1,761	66.2	1,448	0.36	20	29.42	1.58	0.42	6 %
32	1,964	74.3	1,613	0.36	17	30.19	1.81	0.42	6 %
33	2,162	75.2	1,781	0.36	23	31.0	2	0.40	4 %
34	2,345	93.8	1,919	0.34	23	31.87	2.13	0.44	10 %
35	2,503	93.6	2,031	0.37	20	32.28	2.72	0.46	9 %
36	2,624	86.9	2,102	0.37	21	32.66	3.34	0.43	6 %

1. GA, gestational age; EFW, estimated fetal weight; eTFLV, expected total fetal lung volume; OFW: observed fetal weight; O/E TFLV: observed-to-expected total fetal lung volume; %LH: percent liver herniation. ^a Mean EFW (g) published by Salamon et al. ^b Mean expected TFLV (mL) published by Meyers et al. ^c Averages of computer-generated data.

Analysis showed a positive correlation between advancing GA and the difference between standard and adjusted GA, indicating that GA correction increased with GA progression. The corrected O/E TFLV values significantly increased (mean: 43.5 %) compared to standard values (mean: 36.2 %) (p<0.001), highlighting a 7.3 % difference that emphasizes the potential for a significant increase in O/E values with our new approach.

The corrected estimated mortality probability (mean ± SD: 58.6 ± 5.1 %) significantly decreased compared to the standard (mean ± SD: 60.2 ± 4.5 %) (p<0.001), showing an average 1.5 % decrease (shown in Figure 1). Corrected mortality predictions had a lower mean but greater variation, suggesting some patients initially had higher predicted mortality. When grouped by severity, all three groups showed significant differences between standard and corrected predictions (p<0.001 for all), with the mild and moderate groups having the largest differences.

Figure 1:

Comparison of calculated predicted mortality probabilities using a proposed formula by Ruano et al. The corrected predicted mortality exhibited a statistically significant lower mean value with greater variability (p<0.001).

Using standard O/E TFLV and %LH criteria, 30.7 % of infants (n=308) were classified as mild, 48.4 % (n=486) as moderate, and 21 % (n=211) as severe CDH. However, introducing corrected O/E TFLV led to significant re-classifications (Table 2). Specifically, 17.1 % (36/211) initially classified as severe were reclassified as moderate, and 8.6 % (42/486) initially categorized as moderate were reclassified as mild using corrected O/E TFLV.

Table 2:

A comparison of severity classification grading using standard and corrected O/E TFLV.

Frequency (% of 1,005)		Standard severity grade
		Mild	Moderate	Severe	Total
Corrected severity grade	Mild	308	42	0	350 (34.8 %)
	Moderate	0	444	36	480 (47.8 %)
	Severe	0	0	175	175 (17.4 %)
	Total	308 (30.7 %)	486 (48.4 %)	211 (21.0 %)	1,005 (100.0 %)

Mathematical analysis of GA adjustments on the corrected O/E TFLV

Table 3 shows the mean percent differences between corrected and standard O/E TFLV values across all fetal lung volumes from 10 to 65 % O/E TFLV. Generally, 2-week corrections had higher mean percent differences than 1-week corrections for all GAs using Rypens’ and Meyers’ methods. Moreover, the mean percent difference between standard and corrected O/E TFLV decreased as GA increased.

Table 3:

Mean percent differences between the corrected and standard O/E TFLV values by GA using both Rypens’ and Meyers’ formulas.

GA	Rypens correction		Meyers correction
	1-week	2-weeks	1-week	2-weeks
22	5.3 ± 2.3	11.8 ± 5.1	6.1 ± 2.7	13.6 ± 5.9
23	5.1 ± 2.2	11.1 ± 4.8	5.8 ± 2.5	12.9 ± 5.6
24	4.9 ± 2.1	10.6 ± 4.6	5.6 ± 2.4	12.3 ± 5.3
25	4.6 ± 2.0	10.1 ± 4.4	5.3 ± 2.3	11.7 ± 5.1
26	4.5 ± 1.9	9.6 ± 4.2	5.1 ± 2.2	11.2 ± 4.9
27	4.3 ± 1.9	9.2 ± 4.0	4.9 ± 2.1	10.7 ± 4.6
28	4.1 ± 1.8	8.9 ± 3.9	4.7 ± 2.0	10.2 ± 4.4
29	4.0 ± 1.7	8.5 ± 3.7	4.5 ± 2.0	9.8 ± 4.3
30	3.8 ± 1.7	8.2 ± 3.6	4.4 ± 1.9	9.4 ± 4.1
31	3.7 ± 1.6	7.9 ± 3.4	4.2 ± 1.8	9.1 ± 4.0
32	3.6 ± 1.6	7.6 ± 3.3	4.1 ± 1.8	8.8 ± 3.8
33	3.4 ± 1.5	7.3 ± 3.2	3.9 ± 1.7	8.5 ± 3.7
34	3.3 ± 1.5	7.1 ± 3.1	3.8 ± 1.7	8.2 ± 3.6
35	3.2 ± 1.4	6.9 ± 3.0	3.7 ± 1.6	7.9 ± 3.4
36	3.1 ± 1.4	6.7 ± 2.9	3.6 ± 1.6	7.7 ± 3.3

Figure 2 shows the relationship between standard and corrected O/E TFLV across GA (22–36 weeks) for 1-week (dashed lines) and 2-week (solid lines) GA corrections. A horizontal dotted gray line marks the 32 % corrected O/E TFLV threshold, and a vertical dotted gray line shows where it intersects the standard 32 % O/E TFLV. A diagonal dotted black line serves as a reference for no correction.

Figure 2:

A mathematical analysis of GA adjustments by 1- and 2-week corrections on O/E TFLV values between 10 % – 65 % using Rypens’ formula. Corrections include calculation based upon a 1-week (dashed lines) and 2-weeks (solid lines) GA corrections. Black dotted line demonstrates no correction as reference. Gray dashed lines at 32 % demonstrate categorization threshold for severe CDH. Two-week corrections had greater effects across GA compared to 1-week corrections. Positive correlation existed between O/E TFLV and percent difference in values, while GA showed a negative correlation with the percent differences.

Again, we found that percent differences were consistently higher with 2-week corrections (solid lines) compared to 1-week corrections (dashed lines) for each GA. For any specific GA, there was a positive relationship between O/E TFLV and percent difference; as O/E TFLV increased, so did the percent difference between corrected and standard volumes. However, this positive relationship weakened with increasing GA.

Discussion

Accurate assessment of pulmonary hypoplasia in CDH is crucial for prognosis and fetal intervention. Using simulated data, our study explored how adjusting GA norms for fetal weight affects O/E TFLV in isolated CDH and FGR. We hypothesized that FGR-affected fetuses have smaller lung volumes compared to GA-matched counterparts due to compromised overall growth, suggesting that equal lung volumes between FGR-affected and unaffected infants of the same GA may not predict similar outcomes. Therefore, standard methods for O/E TFLV calculation based solely on GA may be misleading in predicting outcomes for these patients.

Our simulation showed a modest but statistically significant 1.5 % decrease in predicted mortality using Ruano’s formula. However, this difference is unlikely to be clinically meaningful, as such a small absolute change in risk would not typically influence decision-making or alter management strategies. Aydin et al. recently reported better survival and lower ECMO need in CDH infants with FGR and an O/E lung-to-head (LHR) ratio<25 %, compared to non-FGR cases [3]. In contrast, prior studies have linked lower survival in CDH neonates treated with ECMO who were<2 kg compared to those>2 kg [12], 13]. However, they did not differentiate outcomes between low birth weight (LBW) infants who were premature vs. those small for gestational age (SGA). Similarly, the CDH Study Group found that LBW infants (<2.5 kg) with CDH had higher rates of morbidity and mortality compared to those>2.5 kg, yet they too did not explore differences in outcomes between appropriate for gestational age (AGA) and SGA infants [14]. Notably, differences in respiratory outcomes between AGA and SGA premature infants have been reported, with some reports that SGA infants suffer less respiratory distress syndrome at birth, albeit with a greater risk of chronic lung disease [15].

Correcting for gestational age in O/E TFLV calculations revealed two important findings. Firstly, the degree of correction matters, with a 2-week correction showing a more significant impact than a 1-week correction, which is intuitive. However, our study was limited by the availability of published FGR data, which extended only to the 3rd percentile, potentially underestimating the impact for infants with severe FGR below the 1st percentile. These infants may experience a delay of up to 4 weeks in GA correction, leading to higher corrected O/E TFLV values.

Secondly, the timing of diagnosis also plays a crucial role, particularly in earlier GA cases, leading to more significant changes in TFLV correction. For example, a growth-restricted infant at 25 weeks’ gestation with a standard O/E TFLV of 26 % would have a 2-week corrected O/E TFLV of 33 % (using Rypens’ formula), categorizing them as having mild or moderate CDH per Ruano et al.’s grading definition. This would theoretically disqualify them from fetal intervention. Conversely, a 32-week gestation infant with the same standard O/E TFLV of 26 % would have a 2-week corrected value of 31 %, maintaining eligibility for FETO in this case. This highlights the critical importance of considering diagnosis timing when assessing eligibility for the FETO procedure in CDH fetuses.

Other imaging parameters for antenatal CDH risk stratification that do not rely on gestational age norms may be useful for infants with CDH and FGR. For instance, in healthy fetuses, TFLV correlates more closely with fetal body volume (FBV) than with gestational age, and percent predicted lung volume (PPLV), calculated by subtracting mediastinal volume from total thoracic volume, predicts CDH severity [16], [17], [18]. Both MR-based measurements account for fetal size, making them relevant for FGR cases. However, both FBV and PPLV are technically challenging and time-consuming to calculate, limiting their widespread use.

While our findings offer a promising theoretical framework, they should not be interpreted as grounds for immediate changes in clinical practice. Current eligibility criteria for fetal interventions such as Fetoscopic Endoluminal Tracheal Occlusion (FETO), as defined by trials like the Tracheal Occlusion To Accelerate Lung growth (TOTAL) trial [19], are based on unadjusted O/E LHR values derived from gestational age norms. Applying a weight-based correction may shift severity classifications, potentially reclassifying some fetuses from severe to moderate. However, it remains uncertain whether these reclassified fetuses would truly experience outcomes consistent with their new severity category. Such reclassification may help avoid unnecessary interventions in some cases, but it also risks denying therapy to those who might still benefit. Given the inherent procedural risks of FETO, including preterm labor, placental abruption, and complications related to balloon placement or removal [20], [21], [22], any adjustments to intervention thresholds must be made with great caution. Our simulation, which examined a broad range of fetal weights, lung volumes, and liver herniation percentages, serves as an important proof of concept. Nonetheless, prospective studies in real-world cohorts are essential to determine whether weight-adjusted metrics improve prognostic accuracy and should be incorporated into clinical decision-making.

As a theoretical proof-of-concept study, our analysis involved several assumptions and inherent limitations. The reference values used for EFW from Salomon et al. were based on healthy GA fetuses without anomalies, potentially limiting their relevance to CDH populations. Additionally, CDH often coexists with congenital anomalies and genetic mutations in 40–60 % of cases, which may influence prognosis and outcome more significantly than lung hypoplasia alone [23], 24]. These complexities, which can contribute to growth restriction and require separate prognostic considerations, were not addressed in our simulation. Furthermore, one critical physiological factor not captured in our simulation is the effect of FGR on lung development itself. FGR has been shown to impair alveolarization, increase pulmonary vascular resistance and arterial stiffness, and contribute to long-term deficits in lung function – all of which could worsen respiratory outcomes regardless of O/E TFLV [25], [26], [27]. These biological effects again underscore the importance of validating our theoretical findings in clinical settings.

An important clinical consideration, though not directly affecting our study, is the challenge of estimating fetal weights in CDH patients. While conventional methods typically rely on abdominal circumference measurements – which may be smaller in CDH-affected fetuses due to displacement of abdominal contents into the thorax – studies have shown that accurate EFW calculations can still be achieved using Hadlock’s formula on prenatal ultrasound [28], 29]. This helps mitigate the difficulties in assessing growth restriction in these cases. Despite these limitations, the core concept of our approach remains: adjusting gestational age based on fetal size in FGR cases results in a higher corrected O/E TFLV than when using standard GA. Nonetheless, the clinical implications of this adjustment must be validated prospectively.

In conclusion, our theoretical simulation highlights the potential impact of incorporating weight-based gestational age adjustments into O/E TFLV calculations, particularly in cases of isolated CDH with FGR. This approach may modestly increase O/E TFLV values, lower estimated mortality, and lead to reclassification within severity categories. While such changes could influence counseling and eligibility for fetal interventions, the observed impact on predicted mortality appears statistically significant yet clinically modest. As such, these findings should be considered hypothesis-generating. Validation through real patient studies is essential to determine the prognostic value and clinical utility of this adjusted approach before it can inform decision-making in practice.