Maternal Intake, BAC, and Weight
Daily alcohol intake (g/kg) for the two sub-strains is shown in (pre- and during-pregnancy exposure), and the daily volume of liquid diet consumption (ml/kg) is shown in . The 4.8% (v/v) alcohol was adopted to reflect a moderate chronic drinking paradigm in humans with consumption generating BAC at an average of 120 mg/dL (4.8%). It has been demonstrated that a liquid diet model with BAC at 80–150mg/dL cause gross retardation and head size reduction. The alcohol level of 4.8% was used to gain additional insight on dose effect on facial dysmorphology. Before pregnancy, the dams of both strains significantly increased their alcohol intake over the first week of exposure [main effect of day on g/kg intake, F(15, 3) = 16.6, p=0.02, and on volume, F(15,3) = 12.2., p=0.03]. However, there were no main or interactive effects of strain for g/kg alcohol intake or volume for the pre-pregnancy period (all p>0.08). BAC ( ) demonstrated a higher significant blood alcohol in B6N compared to B6J in pre-pregnancy with all times examined ( averages, B6N 128 g/dl ± 8, B6J 63 mg/dl ± 6 mg/dl: p<0 .01) In the pregnancy period, the B6N dams drank more than the B6J dams [F(9,9) = 3.3, p=0.05 for g/kg alcohol intake; F(10,8) = 3.52, p=0.04 for volume], associated primarily with higher intakes during the first and last few days of exposure to the alcohol liquid diets during pregnancy (day x line interaction, g/kg F(9,9) = 3.15, p=0.05; volume F(9,9) = 3.48, p=0.04). All p-values are reported in . BAC during pregnancy were considerably reduced in both sub-strains for our time periods of collection with B6N showing significant differences only after 4 hrs into the dark cycle ( averages in pregnancy period C6N 50 ± 5 mg/dl, C6J 21±6 mg/dl: ). For maternal weight gain during pregnancy (E7-E17 weights), there were no significant effects of treatment on sub-strain (). Total number of litters are; Nalc= 7, NPF= 11, NChow= 12 for B6J: and Nalc= 12, NPF= 9, NChow= 7 for B6N.
Figure 1 Alcohol consumption for B6N and B6J animal groups. 1a and b. Alcohol consumption was recorded daily in both B6N and B6J, during pre-pregnancy 16 days (a) and post-pregnancy (b) E7-E17. Daily means -± SEM are reported in g/kg intake. N=30, B6N, (more ...)
Figure 2 A. Chart of blood alcohol concentrations (mg/dl) analized using gas chromatography. Data is reported as MEAN ± SEM with an N=8 for each sun-strain. (*) Denotes significant differences between sub-strain using student t-test. B. Histogram of E17 (more ...)
There were significant treatment effects on weight within each sub-strain revealed for both the B6N and B6J mice [F(2,90)=67.5 and F(2,95)=19.8 respectively; both p<0.0001]. Post-hoc tests for each sub-strain demonstrated significant differences in embryo weights between Alcohol and Chow treatments and between the PF and Chow treatments (all comparisons p<0.0001). Alcohol and PF groups did not differ significantly from each other within either sub-strain (p>0.92). As seen in , within each sub-strain the treatment effects were due to the lower weights of the embryos from the Alcohol and PF dams relative to those from the Chow dams. Moreover, the liquid diet procedure resulted in comparable embryo weights in the Alcohol and PF groups (within sub-strain) on E17. However, embryo weights differed in the two sub-strains in terms of the effects of consuming the control liquid diet. A two-way ANOVA of embryo weights using the Chow and PF groups and the two sub-strains confirmed a significant sub-strain X treatment interaction [F(1,122)=44.9 p<0.0001] as well as significant main effects of treatment [F(1,122)=49.3, p<0.0001] and sub-strain [F(1,122)=10.2, p=0.002]. The Chow group of the B6J sub-strain had higher body weights than the Chow group of the B6N sub-strain (p=0.0001). More importantly, the B6N PF mice (given control liquid diet) had greater reductions in embryo weights (relative to the respective Chow group) than did the B6J mice. Results are based on sample size: NAlc = 37, NPF = 44, NChow = 40 for B6N embryos; and NAlc = 54, NPF = 41, NChow = 28 for B7J embryos. Mean e+ SEM weight for the E17 embryos of the two sub-strains are shown in (B6N, n=121; B6J, n=123).
2D Facial Measurements
The mean and standard errors for each of the 15 facial measurements across treatment groups for each sub-strain are shown in . The only difference between sub-strains for the chow treated animals was in philtrum length (p=0.002). The pair fed animals were different for most of the width measurements (minimal frontal, bitragal, bigonial, inner canthal, outer canthal, whisker pad, columella, all p<0.0001), upper facial depth (p=0.001) and nasal length (p=0.0005). Columella width was larger in the B6N mice but all other differences show larger B6J measurements. The MANOVA models, testing overall treatment effects, yielded significant treatment effects both for the B6N sub-strain (Hotelling’s Lawley-Trace = 2.67, F(30, 70)=3.67, p<0.0001) and the B6J strain (Hotelling’s Lawley-Trace = 3.86, (F30,65)=4.92, p<0.0001). In the B6N sub-strain, significant treatment effects were found for all facial measurements (all p<0.0019) except lower facial depth, philtrum length, and lower facial height (all p>0.20). As shown in , significant differences in the follow-up pair-wise comparisons were observed between the Alcohol and Chow (8 measurements), Alc and PF (8 measurements) and Chow and PF (7- measurements) (all p<0.03). However, four measures were identified in the B6N strain for which the Alcohol group differed significantly both from the PF and the Chow group—upper facial depth (both p<0.03), mid-facial depth (both p<0.008), nasal length (both p<0.0001), and nasal bridge (both p<0.01). In contrast, the B6J strain showed fewer measures with significant treatment effects (whisker pad, upper facial depth, mid-facial depth, and lower facial height, all p<0.002; see ), and only two measures were identified in the B6J strain in which the Alcohol group differed significantly from both the PF and the Chow groups (upper facial depth, and mid-facial depth both p<0.0001). Thus, significant effects specific to alcohol treatment, (i.e., the Alcohol group differed from each respective control group), were identified for two measures of facial depths (upper facial depth; mid-facial depth) in both sub-strains, but only the B6N strain showed significant effects on two other facial height measures (nasal length; nasal bridge). The two strains showed markedly different patterns of facial alterations comparing the PF and Chow groups. In particular, the B6N sub-strain showed a significantly greater vulnerability to the pair feeding procedure (in terms of smaller facial measures) than did the B6J sub-strain. As shown in and , the B6N PF group differed from the Chow group on 7 of the 8 width measures (all p<0.02), whereas the B6J PF group differed from its respective Chow control only on the whisker pad measure (p=0.0001). These changes in embryonic facial measures indicate that the PF feeding procedure had a greater impact on embryonic facial development in the B6N sub-strain than in the B6J sub-strain on E17. This was not simply due to any reduced caloric intake of the B6N PF dams relative to the B6J PF dams during pregnancy, because the B6N dams actually consumed larger volumes of liquid diet (on a ml/kg basis) than did the B6J dams [main effect of strain, F(10,8) = 3.52., p<0.05; strain X day interaction, F(9,9)=3.48, p<0.04]. Interestingly, comparisons between Chow groups (overall line effect across all measurements) demonstrated B6N Chow animals had overall significantly larger measurements for all width measures except palpebral fissure, as well as higher values for nasal length and upper face compared to C6J controls.
Discriminant Analysis of 2D Facial Measures
Discriminant analysis comparing PF to Chow animals in the B6N sub-strain demonstrated perfect identification of the PF treated animals (100% correct) and 92% of the Chow treated animals were correctly classified, using inner canthal, whisker pad, and lower facial depth in the classification function. A similar discriminant analysis model in the B6J sub-strain yielded slightly lower correct clasification rates of 85% for the Chow mice and 87% for the PF mice. Measurements in the final model included outer canthal, whisker pad, upper facial and mid facial depths. While the difference in correct classification of the Chow mice was significant different (p=0.02), there was no significant difference in the correct classification of Chow mice between the two sub-strains (p=0.18). Stepwise discriminant analysis () was performed separately in each sub-strain using the 15 facial measurements, for which mice were identified either as alcohol exposed or non-alcohol exposed (combining Chow and PF mice). For the B6N sub-strain, classification of the alcohol-exposed embryos (n=28) and non-alcohol exposed embryos (n=74) had the highest sensitivity (86%), specificity (80%), and overall correct classification rate (83%) based on one width measurement (bigonial), one depth measurement (lower facial depth), and one height measurement (nasal length). In contrast, the discriminant function for the B6J mice produced the optimal classification of alcohol and non-alcohol exposed embryos using no width measurements, two depth measurements (upper facial, lower facial depth), and one height measurement (nasal bridge). The discriminant analysis for the B6J mice yielded slightly lower significant classification results (sensitivity = 80%, specificity = 78%, overall correct = 79%) than did the discriminant analysis for the B6N mice. The sensitivities between the two sub-strains were not significantly different (p=0.18), and the specificities were marginally different (p=0.06). Additional discriminant analysis comparing alcohol-treated embryos to the PF embryos, yielded similar discriminant functions but with higher correct classification values for both the B6N sub-strain (% overall correct = 90%) and the B6J sub-strain (% overall correct = 85%). Using variables from the final discriminant function (), in conjunction with measurement values, () we show in the representative alcohol induced changes for each sub-strain. The sensitivities between the sub-strains were significantly different (p=0.02), while the specificity were marginally different (p=0.06).Comparison between Chow and PF were also examined for each sub-strain to determine the extent of dysmorphology associated with liquid diet. In B6N, classification of chow (N=34) and PF (n=38) was sensitivity (90%), specificity (85%), and overall correct classification rate (87%) based on 3 measurements; inner canthal and whisker pad widths and lower facial depth. Compared to B6N, classification of B6J required 8 measurement parameters to achieve sensitivity (92%), specificity (100%), and overall correct classification rate (96%) based on outer canthal and whisker pad widths, upper and mid facial depths, and all 4 height measurements (chow, n=19; PF, n=38).
Table 3 Discriminant analysis was done with all 15 measurements for correct predictability. Analysis was performed using backward stepwise model for comparison. Discriminant analysis was performed using individual animal groups (B6N and B6J) with comparisons (more ...)
Figure 4 Embryo photos showing changes in facial measurements used as predictors within the two sub-strains. Red lines arrows indicate the type and direction of change for each sub-strain. In all cases alcohol induced decreased measurements. Calibration is at (more ...)