Abstract
Dear Editor:
Recent literature reviews 5 –7 surveyed clinical studies for comparisons of fetal overgrowth between RI and insulin analogs, including LP, in pregnancies complicated by diabetes. These qualitative reviews concluded that overall there were no relevant differences in rates of neonatal macrosomia and LGA between LP and RI treatment groups. Because of the small numbers of events in each of the studies included in our review, 6 we concluded that there were no relevant differences and presume that the other two reviews may have had the same reasoning, perhaps because each study had small numbers of such events.
The known risk factors that impact infant birth weight are maternal age, body mass index, diabetes duration, parity, infant gender, and smoking status.
8
The meta-analysis by González Blanco et al.
1
noted that baseline characteristics in patients treated with LP were similar to those treated with RI; however, only two factors—age and duration of diabetes—were evaluated in all three studies' data. Availability of baseline data on known risk factors impacting infant birth weight would allow for a more appropriate meta-analysis that properly accounts for these covariates. Ideally, data analysis on the individual patient level would be most informative. It may be useful to provide a brief summary here of the individual observational studies used in the analysis by González Blanco et al.
1
of LGA and LP, all of which are based on a pregnant female patient population with T1DM: • One publication that was not included in the aforementioned literature reviews, an unintentional omission in our review and probably also for the others, was that of Evers et al.,
2
based on a prospective observational study in The Netherlands that included 289 patients, 11% of whom were treated with LP. The study found a high percentage of macrosomia (48.8%), defined as birth weight >90th percentile, despite good glycemic control in the total cohort. The authors concluded an association of LP and LGA on the basis of an odds ratio that was not statistically significant (odds ratio 3.1, 95% confidence interval 0.9–10.4). Evers et al.
2
stated that maternal age, race, and parity were not associated with macrosomia. Additionally, they reported that duration of diabetes, long-term diabetes complications, total daily insulin dosage, body mass index, and weight gain were similar between those groups with macrosomia and without macrosomia. However, in the multivariate model, weight gain did contribute to macrosomia. Nonetheless, the authors remarked that it may have been a select group of women who were prescribed LP, and not the LP itself as the reason for the increased incidence. Our review
6
included a subsequent publication by Evers et al.
9
based on the same study that did not point out any association of LP with LGA or any congenital abnormalities. • The prospective study by Cypryk et al.
3
collected data from a diabetes outpatient clinic in Poland (n=71) that included 25 patients treated with LP and 46 treated with RI. They found 10 (43.5%) cases of LGA (defined as birth weight >90th percentile) in the LP group as opposed to 13 (30.3%) cases of LGA in the RI group. The authors concluded that LGA (defined in this study as birth weight greater than the 90th percentile) did not differ between groups. While the study stated that maternal age, duration of diabetes, severity of diabetes, body mass index, and hemoglobin A1c (HbA1c) were similar between groups, the study did not adjust for these data in the analysis. • Lapolla et al.
4
used Italian-only data from a multicenter, multinational retrospective study (n=370). This study reported similar values for age, duration of diabetes, and baseline HbA1c between treatment groups: LP (n=72) and RI (n=298). As originally noted in the review by Edson et al.,
6
the study found a higher proportion of LGA infants (defined as birth weight >90th percentile) in the LP-treated group compared with the RI group (55.1% [95% confidence interval, 42.6–67.1] vs. 39.2% [95% confidence interval, 29.2–41.0], P=0.0267, respectively). However, differences in macrosomia rates (defined as birth weight >4,000 g) between treatment groups were not significant nor were differences in newborn birth weight. It is interesting to note that González Blanco et al.
1
did not find an association between LP and birth weight. Lapolla et al.
4
noted that the higher frequency of LGA did not seem to correlate with the duration of exposure to LP because no difference was seen in the frequency of LGA infants between patients given LP throughout pregnancy and those stopping LP after the first trimester. The study reported similar maternal age, duration of diabetes, and baseline HbA1c between treatment groups. However, there was no adjustment for these factors in the analysis.
In all pregnancies complicated by diabetes mellitus, the ultimate goal is to maintain normoglycemia in mother and fetus throughout pregnancy. However, increased rates of macrosomia and perinatal morbidity may still occur despite good maternal glycemic control, as measured by HbA1c. 2,9 Thus, studies assessing pregnancy outcomes in patients with diabetes should control not only for treatment type, but for all known risk factors on an individual level before any robust conclusions are drawn. We agree with González Blanco et al. 1 that it is important that the potential issue of LGA rates in pregnancies treated with LP be assessed by others in a more rigorous study design. However, their conclusion that there is “no evidence of a beneficial effect of LP use during pregnancy” may be premature because of the lack of information. In non-pregnant patients with diabetes the benefits of LP in comparison with RI have included better glycemic control and less hypoglycemia. 10,11 Although based on unadjusted analyses, observational studies in pregnant patients also found benefits of LP over RI, with better glycemic control 4,7,12 and less hypoglycemia. 4,7 A fuller assessment of the risks and benefits of LP remains to be studied in pregnant insulin-treated patients with diabetes.