Diabetes Care 1

Suja Padmanabhan,1,2 Vincent W. Lee,2,3 The Association of Falling Insulin Mark Mclean,1,4,5 Neil Athayde,6 CARE/EDUCATION/NUTRITION/PSYCHOSOCIAL CLIN Valeria Lanzarone,7 Qemer Khoshnow,7 Requirements With Maternal Michael J. Peek,8 and N. Wah Cheung1,2 Biomarkers and Placental Dysfunction: A Prospective Study of Women With Pre-existing in https://doi.org/10.2337/dc17-0391

OBJECTIVE To investigate the association of falling insulin requirements (FIR) among women with pre-existing diabetes with adverse obstetric outcomes and maternal bio- markers longitudinally in pregnancy.

RESEARCH DESIGN AND METHODS 1Diabetes and Endocrinology, Westmead Hospital, A multicenter prospective cohort study of 158 women (41 with type 1 diabetes and Sydney, New South Wales, Australia 117 with ) was conducted. Women with FIR of ‡15% from the peak 2Sydney Medical School, University of Sydney, total daily dose after 20 weeks gestation were considered case subjects (n =32). Sydney, New South Wales, Australia 3 The primary outcome was a composite of clinical markers of placental dysfunction Renal Medicine, Westmead Hospital, Sydney, £ New South Wales, Australia (pre-, small for gestational age [ 5th centile], , premature delivery 4Diabetes and Endocrinology, Blacktown Hospital, [<30 weeks], and placental abruption). Maternal circulating angiogenic markers Sydney, New South Wales, Australia (placental growth factor [PlGF] and soluble fms-like tyrosine kinase 1 [sFlt-1]), 5Western Sydney University, Sydney, New South Wales, Australia placental hormones (human placental lactogen, progesterone, and tumor necrosis 6 a Obstetric Medicine, Westmead Hospital, Sydney, factor- ), HbA1c, and creatinine were studied serially during pregnancy. New South Wales, Australia 7Obstetric Medicine, Nepean Hospital, Sydney, RESULTS New South Wales, Australia FIR ‡15% was associated with an increased risk of the composite primary outcome 8College of Medicine, Biology and Environment, (odds ratio [OR] 4.38 [95% CI 1.9–10.3]; P < 0.001), pre-eclampsia (OR 6.76 [95% CI The Australian National University, Canberra, Australian Capital Territory, Australia 2.7–16.7]; P < 0.001), and was more common among women with type 1 diabetes P Corresponding author: Suja Padmanabhan, (36.6 vs. 14.5%; = 0.002). Creatinine was modestly elevated among women with [email protected]. ‡ FIR 15%; however, there was no difference in HbA1c. The ratio of sFlt-1 to PlGF was Received 22 February 2017 and accepted 9 July significantly higher among women with FIR at 25, 30, and 36 weeks, with differences 2017. maintained in the subgroup that developed pre-eclampsia. There was no difference This article contains Supplementary Data online in placental hormones between the groups. at http://care.diabetesjournals.org/lookup/ suppl/doi:10.2337/dc17-0391/-/DC1. CONCLUSIONS © 2017 by the American Diabetes Association. This is the first prospective study to associate FIR with altered expression of placental Readers may use this article as long as the work antiangiogenic factors and pre-eclampsia. FIR is an important clinical sign, among is properly cited, the use is educational and not for profit, and the work is not altered. More infor- women with pre-existing diabetes, that should alert the clinician to investigate un- mation is available at http://www.diabetesjournals derlying placental dysfunction. .org/content/license. Diabetes Care Publish Ahead of Print, published online August 10, 2017 2 Insulin Requirements and Placental Dysfunction Diabetes Care

Although the majority of women with pre- markersaswellashormonesofinsulin Outcomes existing diabetes need increasing doses resistance. As there is currently no universally ac- of insulin with advancing gestation (1), a cepted definition of placental insufficiency, proportion need a reduction in insulin re- RESEARCH DESIGN AND METHODS the primary outcome was defined a priori quirements in late pregnancy, the signifi- FIRST in Pregnancy was conducted at as a composite of adverse clinical out- cance of which remains unclear. Because three tertiary referral hospitals with ded- comes associated with placental dysfunc- insulin resistance is driven by placental- icated diabetes in pregnancy services in tion in the literature (4,14,15). These mediated hormone production, falling in- Sydney, Australia, between June 2013 included pre-eclampsia (defined as per sulin requirements (FIR) are considered to and October 2015. The study was ap- the International Society for the Study be a marker of fetoplacental compromise, proved by the Western Sydney Local of Hypertension in Pregnancy 2014 crite- prompting admission to hospital for ma- Health District Ethics Committee. ria [16]), SGA (birth weight #5th percen- ternal and fetal monitoring and early de- Women with singleton pregnancy tile [17,18]), preterm delivery (,30 livery in some cases (2,3). and a diagnosis of pre-existing type 1, weeks) (19), placental abruption, and still- However, despite the potential impor- type 2, or a new diagnosis of overt diabe- birth (.20 weeks). Presence of one or tance of this clinical sign, only four stud- tes on a 75-g oral tolerance test more of these outcomes in the pregnancy ies have investigated women with FIR in during pregnancy (defined as a fasting was the threshold for attribution of an the literature. We have previously shown, blood glucose level of $7.0 mmol/L [126 abnormal primary outcome. Additionally, through a retrospective cohort study, that mg/dL], a 2-h level of $11.1 mmol/L secondary outcomes were evaluated, in- FIR are associated with pre-eclampsia, [200 mg/dL], or an HbA1c of $6.5% cluding maternal and neonatal clinical small-for-gestational-age (SGA) babies, [48 mmol/mol] [12,13]) were prospec- outcomes, HbA1c, and serum creatinine, and increased admission to the neonatal tively recruited at their first visit to the as well as biomarkers associated with pla- intensivecareunit(NICU)(4).Incontrast, dedicated diabetes in pregnancy clinic. cental dysfunction, including the antian- other studies have not identified an asso- Only women presenting prior to 20 weeks giogenic factor sFlt-1 and proangiogenic ciation with adverse obstetric outcomes and progressing beyond 20 weeks gesta- factor PlGF (11), and hormones of insu- (5–7). Previous studies are limited by tion were included in the study. lin resistance including tumor necrosis retrospective design, differing defini- Information on baseline demograph- factor-a (TNF-a), human placental lacto- tions of FIR, limited representation of ics and medical and obstetric history gen (hPL), and progesterone (20,21). womenwithtype2diabetes,andin- were obtained via a structured question- Laboratory Procedures adequate statistical power for adverse naire at recruitment. Weight in kilograms, Blood and urine were collected from par- outcomes associated with placental blood pressure, insulin dose (total, basal, ticipants atfourtime points:146 1 weeks dysfunction. and prandial), and carbohydrate intake (or first visit if presenting later), 24 6 1 There is increasing evidence for the per meal (quantified in exchanges; 1 ex- weeks, 30 6 1 weeks, and 36 6 1weeks role of placental biomarkers in predicting change = 15 g) were recorded at each re- gestation. Samples from the end of each tri- and diagnosing obstetric conditions asso- view. For women treated with an insulin mester were tested immediately for HbA , ciated with placental dysfunction; how- pump, an upload containing data on av- 1c creatinine, urine albumin-to-creatinine ever, few studies have included women erage basal and prandial insulin dosing ratio (ACR), and protein-to-creatinine with pre-existing diabetes (8–10). An im- and daily carbohydrate intake for the pre- ratio (PCR), and results were made avail- balance between proangiogenic factors, ceding 1 to 2 weeks was recorded. At all able to the treating team. Remaining such as placental growth factor (PlGF), centers, insulin doses were titrated to samples were stored at 280°C and ana- and antiangiogenic factors, such as solu- the same blood glucose target of #5.5 lyzed within 18 months for sFlt-1, PlGF, ble fms-like tyrosine kinase 1 (sFlt-1), are mmol (100 mg/dL) fasting and #7 mmol TNF-a, hPL, and progesterone. Bio- thought to play a key role in diseases as- (126 mg/dL) 2 h postprandial by the treat- markers were analyzed by ELISAs using sociated with placental dysfunction (11). ing team, and intermittent checks of the commercially available kits: sFlt-1, PlGF, FIR among women with pre-existing glucose monitor were performed to ensure and TNF-a (R&D Systems), hPL (DRG Di- diabetes may be considered a clinical man- correlation between self-recorded glucose agnostics), and progesterone (Labor ifestation of the glycemic vascular inter- values and those in the machine. Partici- Diagnostika Nord). The interassay coefficients face between the fetoplacental unit and pants were reviewed at least every four of variation were 11.1, 15.5, 14, 22, and 5.2% the mother; however, no previous studies weeksupto28weeks,everytwoweeks for PlGF, sFlt-1, TNF-a, hPL, and progester- have examined the association between until 36 weeks, and weekly thereafter. one, respectively. FIR and placental biomarkers. FIR was calculated as a percentage from Considering this, we conducted the the peak total insulin dose (PTID), defined as Statistical Analysis Falling Insulin Requirements Study (FIRST), the highest total insulin dose in pregnancy, Statistical analysis was performed using a prospective multicenter cohort study, and the trough total insulin dose (TTID), de- SPSS version 23 and S-PLUS version 8.2. specifically powered to examine the rela- fined as the lowest total insulin dose follow- The primary analysis was conducted tionship between FIR and clinical outcomes ing the peak, using the following formula: using a case-control model by dichoto- fi of placental insuf ciency among women FIR ¼ðPTID-TTIDÞ=PTIDÞ3100%: mizing the cohort into case subjects (de- with pre-existing diabetes. In addition, we fined as women with $15% FIR after investigated the pathophysiology of FIR Care was taken not to include the period 20 weeks gestation) and control subjects through serial examination of placental bio- within 1 week of antenatal steroid (women with ,15% FIR after 20 weeks). markers including pro- and antiangiogenic administration when calculating FIR. The cut point of 15% was chosen, as this care.diabetesjournals.org Padmanabhan and Associates 3

was considered clinically significant in cohort. Caucasian ethnicity was also in- babies or stillbirth was small and did not previous studies (4,5,7). For the purpose creased among women with FIR $15%; differ between the groups. of statistical analysis, women with overt however, this was no longer significant Logistic regression analysis was con- diabetes in pregnancy were reclassified as after correction for type of diabetes by ducted to analyze the independent predic- type 1 diabetes if positive for GAD anti- regression analysis. tors of the primary composite outcome bodies or type 2 diabetes if negative. Dif- Glycemic control, measured by HbA1c and pre-eclampsia. After correction of fac- ferences in baseline characteristics were at the end of each trimester (trimester 1: tors significantly associated with either FIR corrected for type of diabetes by logistic 13 [12–15] weeks; trimester 2: 24 [24–26] or the outcome of interest on univariable regression. Based on results from our pre- weeks; and trimester 3: 36 [35–36] analysis, FIR remained an independent viously published retrospective study, a weeks), improved in both groups with ad- predictor of the primary composite out- sample size of 156 women was calculated vancing gestation; however, there was no come but was not independent of third- to give a power of 80% at a type 1 error of significant difference in HbA1c between trimester creatinine for pre-eclampsia 0.05 for the primary outcome using this women with and without FIR $15%. Like- (Supplementary Tables 1 and 2). definition of FIR (4). wise, there was no difference in gesta- Babies of women with FIR $15% were Continuous variables were analyzed af- tional weight gain and FIR status (Table 1). more likely to be preterm (,37 weeks) ter log transformation using t tests and Serum creatinine was modestly higher and more likely to require emergency cae- summarized as median and lower to up- in women with FIR $15% throughout sarean section and admission to NICU. per quartile (quartile 1 to quartile 3), pregnancy and remained independently There was higher antenatal steroid use whereas Pearson x2 or Fisher exact tests associated with FIR even after correction among women with FIR $15% in keeping were used for categorical variables. for type of diabetes (Table 1). To further with earlier deliveries. Results of the neo- Biomarker values were analyzed longitu- assess the relationship between creati- natal and maternal secondary outcomes dinally during pregnancy by linear mixed- nine and FIR, the change in creatinine are summarized in Table 2. effects models after log transformation. from the start to end of the second and Correction for maternal age, BMI, ethnic- third trimester was calculated (D creati- Placental Biomarkers ity (Caucasian, Asian, or other), nullipar- nine). Women with FIR $15% had a sig- Results for sFlt-1 and PlGF are not pre- ity, and smoking status was performed for nificantly greater increase in creatinine sented, as the ratio is known to be a the analysis of angiogenic markers, as during the third trimester (6.5 [4–12] more sensitive marker of placental dys- previous studies have shown these vari- vs. 4 [22to8]mmol/L; P = 0.044) function (9,23). The sFlt-1/PlGF ratio ables can influence levels (8,22). (Supplementary Fig. 1). There was no dif- was significantly higher among women ference in ACR or PCR between the two with FIR from 25 weeks onwards (Fig. RESULTS groups. 1A). These results remained significant af- One hundred fifty-eight women were ter adjustment for baseline differences in included in the final analysis, of which Clinical Outcomes type of diabetes, as well as factors that 32 (20.3%) fit the case definition of In total, 33 (20.9%) women were positive have previously been shown to lower FIR $15%. The median reduction in insu- for the primary composite outcome. PlGF levels, including maternal age, BMI, lin requirements among the case group Women with FIR $15% had a significantly ethnicity, nulliparity, and smoking status was 25.2% (20.9–43.7%), substantially greater risk of developing the composite (8,22). To account for the effect of pre- greater than the chosen cut point of outcome compared with those without eclampsia on the differences in PlGF and 15%. The percent fall in insulin require- (odds ratio 4.38 [95% CI 1.86–10.3]; P , sFlt-1, the analysis was repeated after ments was similar whether the dose was 0.001) (Table 2). This remained significant separating the cohort into those with calculated in units or corrected for weight after women who had reduced carbohy- and without pre-eclampsia. In total, (units per kilogram). drate intake were excluded (46.2 vs. 27 women developed pre-eclampsia, A comparison of maternal baseline 15.9%; P = 0.001) and when FIR were cal- of whom 14 also had FIR $15%. These and pregnancy characteristics between culated using weight-corrected dose women had significantly higher sFlt-1/ women with and without FIR $15% is (41.7 vs. 14.8%; P = 0.001) and weight- PlGF ratio at 25 and 36 weeks (Fig. 1B). summarizedinTable1.Fortywomen corrected basal dose (36.7 vs. 17.2%; The remaining cohort consisted of 131 (25.3%) had type 1 diabetes, 94 (59.5%) P = 0.018). women, of whom 18 also had FIR $15%. had type 2 diabetes, and 24 (15.2%) were The increased incidence of the com- In this subgroup, there were no longer diagnosed with overt diabetes during posite outcome among women with FIR any significant differences in sFlt-1/PlGF pregnancy, of whom 1 was reclassified was largely driven by the increased risk of ratio between women with and without as type 1 diabetes and 23 as type 2 diabe- pre-eclampsia (odds ratio 6.76 [95% CI FIR (Fig. 1C). tes. A greater proportion of women with 2.74–16.70]; P , 0.001) (Table 2). Appro- As the number of women in each sub- type 1 diabetes developed FIR $15% ximately half of the women with pre- group was small and not powered to compared with type 2 or overt diabe- eclampsia had FIR $15%. Of these show differences between groups, we ex- tes (37.5 vs. 14.9 vs. 12.5%, respectively; women, the majority (n =9)hadevidence amined if there was a trend of increasing P = 0.012). Treatment with aspirin during of FIR preceding the clinical diagnosis of sFlt-1/PlGF ratio across the four sub- pregnancy was higher among women pre-eclampsia by 3 (0–4) weeks (Supple- groups at 36 weeks. We hypothesized with FIR $15%, independent of the type mentary Fig. 2). There were no deliveries that women with no pre-eclampsia and of diabetes, although overall, very few prior to 30 weeks and no episodes of pla- no FIR (group 1) would have the least pla- women were treated with aspirin in this cental abruption. The incidence of SGA cental dysfunction, followed by women 4 Insulin Requirements and Placental Dysfunction Diabetes Care

Table 1—Characteristics of women with and without FIR ‡15% FIR $15% FIR ,15% Variables (case subjects, n = 32) (control subjects, n = 126) P value P value* Age (years) 32.0 (28.5–35.0) 32.5 (29.0–36.0) 0.603 d Prepregnancy BMI (kg/m2) 27.7 (24.2–33.2) 30.45(26.3–38.1) 0.202 d Ethnicity Caucasian 17 (53.1) 40 (31.7) 0.026 0.180 Asian 12 (37.5) 49 (38.9) Other 3 (9.4) 37 (29.4) Diabetes 0.012 d Type 1 15 (46.9) 25 (19.8) Type 2 14 (43.8) 80 (63.5) Overt 3 (9.4) 21 (16.7) Duration of diabetes (years) 5.5 (1.25–15.5) 3 (1–8) 0.134 d Microvascular complications 7 (21.9) 19 (15.1) 0.35 d Nulliparity 13 (40.6) 37 (29.4) 0.221 d Unplanned pregnancy 17 (53.1) 76 (60.3) 0.460 d Chronic hypertension 6 (18.8) 13 (10.3) 0.223 d Hypertensive disorder in previous pregnancy† 5 (26.3) 22 (24.7) 0.805 d Current smoker 4 (12.5) 11 (8.7) 0.507 d Aspirin use during pregnancy 5 (21.9) 10 (6.3) 0.014 0.010

Preconception HbA1c (%) 7.4 (6.7–9.4) 7.25 (6.1–9.8) 0.628 d (mmol/mol) 57 (50–79) 56 (43–84)

Trimester 1 HbA1c (%) 6.9 (6.4–7.7) 6.4 (5.8–7.6) 0.132 (mmol/mol) 52 (46–61) 46 (40–60)

Trimester 2 HbA1c (%) 6.2 (5.5–6.8) 5.7 (5.2–6.7) 0.380 (mmol/mol) 44 (37–51) 39 (33–50)

Trimester 3 HbA1c (%) 6.2 (5.3–7.1) 6.1 (5.7–7.3) 0.87 (mmol/mol) 44 (34–54) 43 (39–56) Creatinine (mmol/L) Trimester 1 46.5 (42–54) 43 (39–48) 0.001 0.015 Trimester 2 46 (40–52) 42 (38–47) 0.043 0.053 Trimester 3 57 (44–63) 46 (40–51) ,0.001 0.001 ACR d Trimester 1 0.85 (0–1.7) 0.9 (0–2.1) 0.780 Trimester 2 1.1 (0–1.8) 0.9 (0–1.7) 0.263 Trimester 3 1 (0.6–11.1) 1.8 (0.8–3.6) 0.244 PCR d Trimester 1 13 (0–17) 11 (5.9–16) 0.094 Trimester 2 12.5 (8–21) 16 (9–22.9) 0.717 Trimester 3 21 (15–60) 23 (16–34.5) 0.145 Total gestational weight gain (kg) 14.4 (8–18.3) 12.6 (7.5–17.6) 0.860 d Data represent median (quartile 1 to quartile 3) or n (%) as applicable. *P value correcting for type of diabetes with logistic regression. †Only women with parity $1 were included in this analysis (n =108).

with no pre-eclampsia and FIR (group 2), the third trimester among women with Furthermore, babies of women with thenwomenwithpre-eclampsiaandno FIR $15% (Supplementary Table 3). FIR $15% were more likely to be deliv- FIR (group 3), and women with pre- ered early by emergency caesarean sec- eclampsia and FIR (group 4) would have CONCLUSIONS tion and admitted to the NICU. Although the greatest placental dysfunction. There FIR are considered a marker of placental no specific difference in individual neona- was a significant trend for increasing insufficiency, although to date, few stud- tal outcomes could be detected, it is likely sFlt-1/PlGF ratio from groups 1 to 4, re- ies have supported this theory. In this the higher need for NICU intervention re- spectively (Fig. 2). study, consistent with our previously pub- flects preterm delivery, pre-eclampsia, There was no difference in TNF-a,hPL, lished retrospective analysis (4), we found and general poorer condition of babies and progesterone levels between women that FIR of $15% increased the risk of in the FIR group, suggesting poorer pla- with and without FIR at any of the four having an adverse outcome associated cental health overall. gestational windows examined longitudi- with placental dysfunction by more than Although the underlying origin of pla- nally during pregnancy, nor was there a fourfold, driven primarily by an increased cental dysfunction has not been clearly greater reduction in hormone levels over risk of pre-eclampsia of more than sixfold. identified, it is accepted that it causes a care.diabetesjournals.org Padmanabhan and Associates 5

Table 2—Maternal and neonatal outcomes among women with and without FIR ‡15% FIR $15% FIR ,15% Clinical outcome (case subjects, n = 32) (control subjects, n = 126) P value Neonatal outcomes Any adverse neonatal outcome† 25 (78.1) 90 (71.4) 0.447 Gestational age of delivery (weeks) 37.4 (35.7–38.1) 38.1 (37.6–38.6) 0.022* Preterm delivery ,30 weeks 0 (0) 0 (0) d ,37 weeks 14 (43.8) 23 (18.3) 0.002* Stillbirth 0 (0) 1 (0.8) 1.00 Admission to NICU‡ 12 (37.5) 26 (20.8) 0.049* Birth weight 2,992.5 (2,830–3,635) 3,465 (3,085–3,865) 0.043* Birth weight centile§ 47.73 (20.6–94.77) 68.63 (42.13–93.46) 0.308 Large for gestational age ($90th centile)§ 9 (28.1) 37 (29.4) 0.890 Small for gestational age #5th centile§ 1 (3.1) 8 (6.3) 0.688 #10th centile§ 4 (12.5) 9 (7.1) 0.301 Jaundice‡ 12 (37.5) 38 (30.4) 0.442 ‡ 18 (56.2) 66 (52.8) 0.727 4 (12.5) 11 (11.2) 0.764 Congenital malformation (major and minor)‡ 6 (18.8) 13 (10.4) 0.225 Respiratory distress‡ 11 (34.4) 37 (29.6) 0.601 Maternal outcomes Primary composite outcome 14 (43.8) 19 (15.1) ,0.001* Pre-eclampsia 14 (43.8)| 13 (10.3)| ,0.001* Gestational hypertension 1 (3.1) 7 (5.6) 1.00 Placental abruption 0 (0) 0 (0) d Induction of labor 27 (84.4) 98 (77.8) 0.412 Mode of delivery 0.051 Vaginal 6 (18.8) 39 (31) 0.172 Instrumental 1 (3.1) 4 (3.2) 1.00 Elective caesarean 7 (21.9) 45 (35.7) 0.137 Emergency caesarean 18 (56.2) 38 (30.2) 0.006* Antenatal steroids 9(28.1) 9(7.1) 0.003 *P value remained significant after adjustment for type of diabetes (1 or 2) by logistic regression. †Composite of stillbirth, admission to NICU, hypoglycemia, jaundice, birth trauma, congenital malformation, respiratory distress, and neonatal death. ‡Stillbirth excluded. §Customized birth weight centile calculated using the gestation network Australian centile calculator (correcting for gestational age, maternal height and weight at booking visit, ethnicity, parity, and fetal sex [40]). |Three women in the case group and one woman in the control group were diagnosed with pre-eclampsia, in the absence of proteinuria, through the criteria of hypertension and organ dysfunction as per the International Society for the Study of Hypertension in Pregnancy (2014) definition (16). spectrum of obstetric disorders, including to delivery in a retrospective study of even after excluding women with reduction placental abruption, pre-eclampsia, and 54 women with type 1 diabetes, whereas in carbohydrate intake and applying weight- ultimately intrauterine growth restriction the 5-year retrospective study by McManus adjusted basal and total insulin dose, our if maternal–fetal exchange is impaired and Ryan (5) examined weight-adjusted association with the primary outcome was (14,24,25). We hypothesize that FIR is an- insulin requirements in 51 maintained. It is likely that previous negative other such clinical expression of this spec- progressing beyond 36 weeks only. findings are because of inadequate power trum of disease, manifesting uniquely Both studies used a case definition of for differences in outcomes, and indeed, this among women with diabetes as a marker FIR $15% in keeping with our study. The study is the first to apply a prespecified cal- of decreasing insulin resistance in the set- third study examining 236 women with culated sample size. ting of a failing . The strength of type 1 diabetes by Steel et al. (6) calcu- Importantly, we found that FIR either the association demonstrated with pre- lated FIR using the PTID and TTID, similar preceded or accompanied the onset of eclampsia supports this hypothesis and to us, but defined the case group as hav- pre-eclampsia in all but three cases, fur- suggests a shared pathology contributing ing FIR $30%. ther supporting our hypothesis that FIR to both outcomes. As yet, there is no clear definition of may be a milder clinical or early manifes- Our findings are in contrast to three the level and timing at which falling re- tation of placental dysfunction. The tem- previous studies on FIR, which did not quirements become clinically significant. poral association of FIR and outcomes find an association with pre-eclampsia Further, the possibility of FIR being a man- was examined in the study by Steel or other adverse outcomes. A potential ifestation of reduced carbohydrate intake et al. (6), who also reported FIR preceded reason for the disparate findings were rather than true changes in endogenous the onset of hypertension and pre- the different definitions used. Achong needs has been the rationale for using eclampsia in the two case subjects deliv- et al. (7) reported weight-adjusted basal basal- and weight-based changes in insu- ered for this indication. Our findings thus insulin requirements from 30 weeks lin dose in previous studies. However, open the possibility of using FIR as a 6 Insulin Requirements and Placental Dysfunction Diabetes Care

there is little evidence for the association between maternal biomarkers and reduc- tions in insulin dose. This is the first pro- spective study to demonstrate altered levels of placental hormones involved in angiogenesis, with a higher ratio of sFlt-1/PlGF among women with FIR from 25 weeks onwards. PlGF is a vascular en- dothelial growth factor homolog, which is released in increasing amounts by the pla- centa during pregnancy and plays a key role in maintaining vascular homeostasis during placental development (26). In contrast, sFlt-1 acts as a negative regula- tor of angiogenesis and has been found in increased amounts in conditions associ- ated with placental dysfunction including pre-eclampsia (27,28). Very few studies have examined pro- and antiangiogenic factors among women with diabetes longitudinally during preg- nancy. Yu et al. (29) demonstrated that women with type 1 diabetes and pre- eclampsia had increased sFlt-1, decreased PlGF, and increased sFlt-1/PlGF ratio com- pared with non–pre-eclamptic women. Another study by Cohen et al. (30) includ- ing women with both type 1 and type 2 diabetes showed similar results. Although neither of these studies documented changes in insulin requirements, their findings mirror those we observed be- tween women with and without FIR. As pre-eclampsia is well known to be associated with derangement of angio- genic factors, a subgroup analysis was conducted to evaluate whether the higher sFlt-1/PlGF ratio among women with FIR was because of the increased in- cidence of pre-eclampsia in this group. Despite small numbers, the differences in sFlt-1/PlGF ratio between women with and without FIR were maintained in the subgroup that developed pre- eclampsia signifying that differences in angiogenic factors cannot be explained by pre-eclampsia alone. Studies have sup- ported the imbalance of antiangiogenic to proangiogenic factors, released by the placenta as a plausible mechanism for the endothelial dysfunction seen in pre- Figure 1—Longitudinal differences in sFlt-1/PlGF ratio between women with FIR $15% (case subjects, dashed lines) and FIR ,15% (control subjects, solid lines) during pregnancy in the total cohort (A), eclampsia (31). More recently, extracor- subgroup with pre-eclampsia (B), and subgroup without pre-eclampsia (C). Markers represent median poreal removal of sFlt-1 and infusion of biomarker levels, and error bars represent interquartile range. y-axis is on log10 scale. *P , 0.05; **P , PlGF has led to improvement in pre- 0.01 for comparisons at each gestational window conducted on log-transformed data using linear mixed- eclampsia, further supporting the causal effects models adjusted for type of diabetes, maternal age, BMI, ethnicity, nulliparity, and smoking status. relationship (32,33). The wide clinical spectrum of severity in pre-eclampsia predictor of pre-eclampsia or a clinical red Historically, the mechanism of FIR has has been well described and is propor- flag to alert clinicians to increase monitor- been attributed to reduction in hormones tional to maternal levels of pro- and anti- ing for this condition. mediating insulin resistance; however, angiogenic factors from previous studies care.diabetesjournals.org Padmanabhan and Associates 7

despite this, women with FIR $15% were still sixfold more likely to develop pre- eclampsia, further reinforcing the strength of the association. Further, the titration of insulin requirements was subjective and may have varied by clinician. Ideally, an objective measure of insulin sensitivity would be favored; however, techniques such as insulin clamps and HOMA are less practical. Finally, the analysis of biomarkers at the prespecified gesta- tions may have missed abrupt or sudden changes in hormone levels, and we were unable to measure the entire spectrum of insulin resistance placental hormones. Thus, it is important for our findings to be validated in larger adequately powered studies before discounting the contribu- tion of placental hormones to changes in insulin requirements. Nevertheless, this is the first study spe- cifically designed to answer this impor- tant clinical question using pilot data to ensure adequate power in detecting dif- Figure 2—Trend of logarithmically transformed angiogenic biomarkers across four subgroups of in- ferences in outcomes. Examination of creasing placental dysfunction. Markers represent mean, and error bars represent 95% CI for logarith- maternal biomarkers, particularly those mically transformed sFlt-1/PlGF ratio. FIR $15%. r is Spearman rank correlation coefficient. P value is associated with placental dysfunction, for Spearman rank correlation across four subgroups with no pre-eclampsia and no FIR considered the group with least placental dysfunction and pre-eclampsia and FIR considered the group with is a novel method of exploring the mech- greatest placental dysfunction. Only those with blood available at 36 weeks included in the analysis. anism of FIR, which has not been reported before. The demonstrated changes in pro- and (31,34,35). In keeping with this, our find- GFR has been associated with a 30–50% antiangiogenic markers, together with ings suggest that women with pre- reduction in insulin requirements (38,39); the association with pre-eclampsia, have eclampsia who manifest FIR may lie on therefore, it is plausible that reductions in important clinical implications, as they the spectrum of more severe placental the renal clearance of insulin results in FIR suggest FIR are a manifestation of placen- dysfunction, as demonstrated by higher in some women during pregnancy, al- tal dysfunction lying on a similar spectrum sFlt-1/PlGF ratio. thoughthedegreetowhichthismay to the other diseases. It is likely that FIR In contrast, we did not find any differ- have affected our results is unclear. It is are an early indicator of placental dys- ence in hPL, progesterone, and TNF-a be- possible that vascular dysfunction is a uni- function in women with diabetes, with tween the groups. Limited studies have fying mechanism for FIR manifesting only a proportion of women subse- examined changes in placental hormones through renal and placental pathways quently progressing to the more serious and insulin requirements and have not and is consistent with the strong associa- clinical manifestations that have typically demonstrated a correlation between hor- tion with pre-eclampsia and disruption in been associated with disrupted angio- mone levels and insulin dose (36,37). In angiogenic markers in our study. How- genic markers. Importantly, our data are keeping with this, we were unable to ever, the similar rates of albuminuria also reassuring in that over half of the show any significant differences between and proteinuria between groups does women with FIR did not develop adverse women with and without FIR despite ex- not correspond with this hypothesis, outcomes or have babies requiring amining absolute and relative changes in and we recognize that using serum creat- NICU admission. Thus, although these hormone levels. inine as a surrogate for renal function in findings support the clinical recommen- In contrast, we detected a significant pregnancy is far from ideal. Further stud- dation that all women manifesting FIR association between higher serum creat- ies are needed to validate these prelimi- ($15%) should have increased surveil- inine level per trimester as well as rising nary findings. lance and investigation for adverse ob- creatinine in third trimester with FIR dur- We acknowledge our study has some stetric outcomes, its presence does not ing pregnancy. Although the absolute limitations. Aspirin was not universally necessarily indicate urgent, immediate difference is small, and the clinical impli- commenced in all women, and a small delivery. cation of these findings is unclear, the but significantly greater number were association raises the possibility that there treated with aspirin in the FIR $15% Acknowledgments. The authors thank Trudi is a renal contribution to the mechanism group. Aspirin is known to reduce the Jones (Blacktown Hospital, Sydney, New South forFIR,inadditiontoplacentaldysfunc- risk of pre-eclampsia and, as a result, Wales, Australia) for technical support in completing tion. Outside of pregnancy, a declining could introduce a negative bias; however, the ELISA assays for this study and the. 8 Insulin Requirements and Placental Dysfunction Diabetes Care

Funding. The authors thank Westmead Medical with suspected preeclampsia. N Engl J Med origins of placental insufficiency. Semin Fetal Neo- Research Foundation for funding the study. 2016;374:13–22 natal Med 2004;9:357–369 Duality of Interest. No potential conflicts of in- 11. Herraiz I, Simon´ E, Gomez-Arriaga´ PI, et al. 26. Ferrara N, Davis-Smyth T. The biology of vas- terest relevant to this article were reported. Angiogenesis-Related Biomarkers (sFlt-1/PLGF) cular endothelial growth factor. Endocr Rev 1997; Author Contributions. S.P. designed the study, in the Prediction and Diagnosis of Placental Dys- 18:4–25 collected and analyzed the data, and wrote the function: An Approach for Clinical Integration. Int 27. Karumanchi SA. Angiogenic factors in pre- manuscript. V.W.L. and N.W.C. designed the J Mol Sci 2015;16:19009–19026 eclampsia: from diagnosis to therapy. Hyperten- study and reviewed and edited the manuscript. 12. Metzger BE, Gabbe SG, Persson B, et al.; In- sion 2016;67:1072–1079 M.M. and M.J.P. reviewed and edited the man- ternational Association of Diabetes and Pregnancy 28. Karumanchi SA, Lindheimer MD. Preeclamp- uscript. N.A., V.L., and Q.K. researched data. All Study Groups Consensus Panel. International asso- sia pathogenesis: “triple a rating”-autoantibodies participating investigators in each center have ciation of diabetes and pregnancy study groups and antiangiogenic factors. Hypertension 2008; been included as authors on this paper. S.P. is the recommendations on the diagnosis and classifica- 51:991–992 guarantor of this work and, as such, had full access tion of hyperglycemia in pregnancy. Diabetes Care 29. Yu Y, Jenkins AJ, Nankervis AJ, et al. Anti- to all the data in the study and takes responsibility 2010;33:676–682 angiogenic factors and pre-eclampsia in type for theintegrityof thedataand theaccuracy of the 13. World Health Organization. Diagnostic criteria 1 diabetic women. Diabetologia 2009;52:160– data analysis. and classification of hyperglycaemia first detected 168 Prior Presentation. Preliminary data from this in pregnancy [article online], 2013. Available from: 30. Cohen AL, Wenger JB, James-Todd T, et al. studywerepresentedinoralformattheAustralian http://apps.who.int/iris/bitstream/10665/85975/ The association of circulating angiogenic factors Diabetes in Pregnancy Society Annual Scientific 1/WHO_NMH_MND_13.2_eng.pdf. Accessed and HbA1c with the risk of preeclampsia in Meeting, Gold Coast, Queensland, Australia, 22 February 2017 women with preexisting diabetes. Hypertens 26–27 August 2016. 14. Baschat AA. Pathophysiology of fetal growth Pregnancy 2014;33:81–92 restriction: implications for diagnosis and surveil- 31. Levine RJ, Maynard SE, Qian C, et al. Circulat- References lance. Obstet Gynecol Surv 2004;59:617–627 ing angiogenic factors and the risk of preeclamp- 1. Padmanabhan S, Jiang S, Mclean M, Cheung 15. Costa SL, Proctor L, Dodd JM, et al. Screening sia. N Engl J Med 2004;350:672–683 NW. Effect of pregnancy on insulin requirements for placental insufficiency in high-risk pregnan- 32. Thadhani R, Hagmann H, Schaarschmidt W, differs between type 1 and type 2 diabetes: A cies: is earlier better? Placenta 2008;29:1034– et al. Removal of soluble Fms-like tyrosine kinase- cohort study of 222 pregnancies. Aust N Z J Obstet 1040 1 by dextran sulfate apheresis in preeclampsia. Gynaecol 2016;56:352–357 16. Tranquilli AL, Dekker G, Magee L, et al. The J Am Soc Nephrol 2016;27:903–913 2. Jovanovic L, Druzin M, Peterson CM. Effect of classification, diagnosis and management of the 33. Spradley FT, Tan AY, Joo WS, et al. Placental euglycemia on the outcome of pregnancy in hypertensive disorders of pregnancy: A revised growth factor administration abolishes placental insulin-dependent diabetic women as compared statement from the ISSHP. Pregnancy Hypertens ischemia-induced hypertension. Hypertension with normal control subjects. Am J Med 1981;71: 2014;4:97–104 2016;67:740–747 921–927 17. Platz E, Newman R. Diagnosis of IUGR: tradi- 34. Wikstrom¨ A-K, Larsson A, Eriksson UJ, Nash P, 3. Essex N. Management of maternal diabetes in tional biometry. Semin Perinatol 2008;32:140– Norden-Lindeberg´ S, Olovsson M. Placental pregnancy. In Carbohydrate Metabolism in Preg- 147 growth factor and soluble FMS-like tyrosine nancy and the Newborn. 363. Sutherland H, 18. Alahakoon TI, Zhang W, Trudinger BJ, Lee VW. kinase-1 in early-onset and late-onset preeclamp- Stowers J, Eds. New York, Springer-Verlag, 1979 Discordant clinical presentations of preeclampsia sia. Obstet Gynecol 2007;109:1368–1374 4. Padmanabhan S, McLean M, Cheung NW. and intrauterine fetal growth restriction with sim- 35. Pinheiro CC, Rayol P, Gozzani L, et al. The re- Falling insulin requirements are associated with ilar pro- and anti-angiogenic profiles. J Matern lationship of angiogenic factors to maternal and adverse obstetric outcomes in women with pre- Fetal Neonatal Med 2014;27:1854–1859 neonatal manifestations of early-onset and late- existing diabetes. Diabetes Care 2014;37:2685– 19. Henderson-Smart DJ, Hutchinson JL, onset preeclampsia. Prenat Diagn 2014;34:1084– 2692 Donoghue DA, Evans NJ, Simpson JM, Wright I; 1092 5. McManus RM, Ryan EA. Insulin requirements Australian and New Zealand Neonatal Network. 36. Spellacy WN, Cohn JE. Human placental lac- in insulin-dependent and insulin-requiring GDM Prenatal predictors of chronic lung disease in togen levels and daily insulin requirements in women during final month of pregnancy. Diabe- very preterm infants. Arch Dis Child Fetal Neona- patients with diabetes mellitus complicating preg- tes Care 1992;15:1323–1327 tal Ed 2006;91:F40–F45 nancy. Obstet Gynecol 1973;42:330–333 6. Steel JM, Johnstone FD, Hume R, Mao JH. In- 20. Kirwan JP, Hauguel-De Mouzon S, Lepercq J, 37. Stewart MO, Whittaker PG, Persson B, sulin requirements during pregnancy in women et al. TNF-alpha is a predictor of insulin resistance Hanson U, Lind T. A longitudinal study of circulat- with type I diabetes. Obstet Gynecol 1994;83: inhumanpregnancy. Diabetes 2002;51:2207–2213 ing progesterone, oestradiol, hCG and hPL during 253–258 21. Ryan EA, Enns L. Role of gestational hor- pregnancy in type 1 diabetic mothers. Br J Obstet 7. Achong N, Callaway L, d’Emden M, McIntyre mones in the induction of insulin resistance. Gynaecol 1989;96:415–423 HD, Lust K, Barrett HL. Insulin requirements in late J Clin Endocrinol Metab 1988;67:341–347 38. Kulozik F, Hasslacher C. Insulin requirements pregnancy in women with type 1 diabetes melli- 22. Akolekar R, Zaragoza E, Poon LC, Pepes S, in patients with diabetes and declining kidney tus: a retrospective review. Diabetes Res Clin Nicolaides KH. Maternal serum placental growth function: differences between insulin analogues Pract 2012;98:414–421 factor at 11 + 0 to 13 + 6 weeks of gestation in the and human insulin? Ther Adv Endocrinol Metab 8. Poon LC, Nicolaides KH. First-trimester mater- prediction of pre-eclampsia. Ultrasound Obstet 2013;4:113–121 nal factors and biomarker screening for pre- Gynecol 2008;32:732–739 39. Biesenbach G, Raml A, Schmekal B, eclampsia. Prenat Diagn 2014;34:618–627 23. Rana S, Powe CE, Salahuddin S, et al. Angio- Eichbauer-Sturm G. Decreased insulin require- 9. Verlohren S, Herraiz I, Lapaire O, et al. The sFlt- genic factors and the risk of adverse outcomes in ment in relation to GFR in nephropathic type 1 1/PlGF ratio in different types of hypertensive women with suspected preeclampsia. Circulation and insulin-treated type 2 diabetic patients. Dia- pregnancy disorders and its prognostic potential 2012;125:911–919 bet Med 2003;20:642–645 in preeclamptic patients. Am J Obstet Gynecol 24. Weissgerber TL, Mudd LM. Preeclampsia and 40. Gardosi J, Francis A. A customised Weight 2012;206:58.e1–58.e8 diabetes. Curr Diab Rep 2015;15:9 Centile Calculator: Gestation Network [Internet], 10.ZeislerH,LlurbaE,ChantraineF,etal.Pre- 25. Chaddha V, Viero S, Huppertz B, Kingdom J. 2013. Available from http://www.gestation.net/. dictive value of the sFlt-1:PlGF ratio in women Developmental biology of the placenta and the Accessed 22 February 2017