All babies who were defined as having acquired early-onset group B streptococcal sepsis had a positive cultureresult for group B streptococci. Other infants with signsof sepsis but a negative result were not included in thestudy. It may be that the incidence of early-onset group Bstreptococcal sepsis is higher than what was estimated inother studies, but we think that our procedure of includ-ing only babies with clinical symptoms plus positiveblood cultures is rigorous enough.

Erich Hafner, MDGynakologische und Geburtshilfliche, Abteilung Donauspital am SMZ-Ost, Langobardenstraβe 122, A-1220, Vienna, Austria

6/8/98124

How to increase the proportion of vaginal hysterec-tomies—Bipolar coagulationTo the Editors: I read with great interest the article byDavies et al (Davies A, Vizza E, Bournas N, O’Connor H,Magos A. How to increase the proportion of hysterec-tomies performecl vaginally. Am J Obstet Gynecol1998:179:1008-12). As do many gynecologists, I share theopinion of the authors that the vaginal route is preferredwhenever possible when a hysterectomy is performed. Iagree with the usefulness of better training in vaginaltechniques, special curved clamps, endoloop sutures,and transvaginal endoscopic oophorectomy. I also sug-gest the use of bipolar coagulation instead of sutures.This is a simple and efficient technique for desiccation ofvessels such as cervical, uterine, and ovarian pedicles.

Between 1992 and 1998, a total of 150 vaginal hys-terectomies with bipolar coagulation (without sutures)has been performed in my unit, without any complica-tions caused by bipolar coagulation or hemostasis. Themain precaution when bipolar coagulation of large pedi-cles (causing temperature elevation of surrounding tis-sues) is performed is to protect bladder and terminalureters with an anterior retractor.

This technique has many advantages. Deep suturesthrough a narrow operating field are avoided, which isuseful for an enlarged uterus or when the cervix remainshigh in the pelvis. With adequate retractors it is almost al-ways possible to get a clear exposition of pedicles, andsimple oophorectomy is most often possible.Laparoscopic exploration is seldom useful and can beavoided in most cases. In a large uterus with large my-omas, desiccation of cervical and uterine pedicles is easyto perform; in rare cases, when bisection, cervical ampu-tation, morcellation, and myomectomy do not allow com-plete treatment by the vaginal route, it is easy to performa laparotomy to complete the excision, because occlu-sion of the main pedicles has already beem realized.

According to my experience bipolar coagulation canincrease the proportion of vaginal hysterectomies.

Bernard P. WeylDepartment of Obstetrics and Gynecology, Centre Hospitalier, 35403 StMalo Cedex, France

6/8/100139

Insulin resistance and preeclampsia: Is there a rela-tionship?To the Editors: With great interest we read the article byJoffe et al (Joffe GM, Esterlitz JR, Levine RJ, Clemens JD,Ewell MG, Sibai BM, Catalano PM. The relationship be-tween abnormal glucose tolerance and hypertensive dis-orders of pregnancy in healthy nulliparous women. Am JObstet Gynecol 1998;179:1032-7). They demonstrated, ina large group of pregnant women, that the 1-hour glu-cose level after a 50-g glucose challenge is a good predic-tor of preeclampsia but not of pregnancy-associated hy-pertension; they concluded that increased insulinresistance may play a role in the pathogenesis of this dis-ease. This is a captivating theory; however, the data pre-sented do not strongly support the conclusion. In fact,the 1-hour glucose level after a 50-g glucose challenge is avery weak indicator of insulin resistance or insulin resis-tance syndrome. Further, given the relationship betweenlevel of plasma glucose and preeclampsia, the findingthat women with abnormal glucose metabolism are notat increased risk of preeclampsia makes these resultseven more difficult to interpret. It is interesting to notethat, in a similar work, Solomon et al1 found quite differ-ent results, with a significantly higher incidence of ele-vated glucose concentrations 1 hour after a 50-g oral glu-cose load in women with gestational hypertension, butnot in those with preeclampsia, when compared withnormotensive pregnant women.

Recently, studies have been designed to evaluate in-sulin sensitivity in women with preeclampsia. Roberts etal2 found that insulin sensitivity, determined with theminimal model technique, was higher in patients withpreeclampsia than in normotensive pregnancies.Consistently, women with preeclampsia had lower fastinginsulin concentrations than normotensive subjects did.In our own study3 of pregnant women with gestationalhypertension and preeclampsia, only the former werecharacterized by metabolic features similar to those ofpatients with insulin resistance syndrome, such as lowerinsulin sensitivity index, higher triglyceride levels, andnonestefified fatty acids. In contrast, we did not observean increased insulin sensitivity index or other metabolicabnormalities in subjects with preeclampsia comparedwith normotensive women.

To enhance the explanation of their findings, Joffe etal should provide information about insulin levels andlipid profile. It would be of interest to evaluate whethertheir patients have the metabolic characteristics of in-sulin resistance syndrome.

Alessandro Caruso, MD, Sergio Ferrazzani, MD, andGiancarlo Paradisi,MD

Department of Obstetrics and Gynecology, Catholic University, Largo A.Gemelli 8, 00168 Rome, Italy

REFERENCES

1. Solomon CG, Graves SW, Greene MF, Seely EW. Glucose intol-erance as a predictor of hypertension in pregnancy.Hypertension 1994;23:717-21.

2. Roberts RN, Henfiksen JE, Hadden DR. Insulin sensitivity inpre-eclampsia. Br J Obstet Gynaecol 1998;105:1095-100.

768 Letters September 1999Am J Obstet Gynecol

3. Caruso A, Ferrazzani S, De Carolis S, Lucchese A, Lanzone A,De Santis L, et al. Gestational hypertension but not preeclamp-sia is associated with the insulin resistance syndrome character-istics. Hum Reprod. In press 1999.

6/8/100142

ReplyTo the Editors: We thank Caruso et al for their interest inour article. However, we do not believe that the data theyquote strongly support their position that increased in-sulin resistance does not play a role in the pathogenesisof hypertensive disorders of pregnancy.

Caruso et al state that the 1-hour glucose level aftera 50-g glucose challenge is a very weak indicator of in-sulin resistance. However, on the basis of our own un-published data in pregnant women, the 1-hour glucosevalue after a 100-g oral glucose tolerance test washighly correlated with insulin resistance as estimatedby the hyperinsulinemic-euglycemic clamp (r =.60; P = .001).

Caruso et al went on to state that women with abnor-mal glucose tolerance are not at risk for preeclampsia.This is true when one only looks at the 6% of women withabnormal glucose tolerance; the odds ratio for develop-ment of preeclampsia was not significant. However, whenwe examined the risk of preeclampsia using quartiles ofplasma glucose after the 50-g glucose challenge in allstudy subjects (n = 3676), there was a significantly in-creased risk of preeclampsia from the second quartilethrough the fourth quartile (odds ratio, 1.58-1.81; P <.028 to .002).

Caruso et al quoted the study of Solomon et al1 toshow that an elevated 1-hour glucose screen was signifi-cantly related to an increased odds ratio of hypertensionin pregnancy, specifically, transient hypertension but notpreeclampsia. Could this not be related to a β error sincethere were only 50 subjects with preeclampsia and 47subjects with transient hypertension when the mean glu-cose value for the normotensive group was 6.0, for thosewith preeclampsia 6.5, and for those with transient hy-pertension 6.9 mmol/L?

Caruso et al also quoted a paper by Roberts et al2showing that insulin sensitivity in women with develop-ment of preeclampsia was actually greater than that inthe control group with a minimal model technique. Itshould be noted that the study group consisted of only 11women. Furthermore, for the minimal model in thisstudy, only an intravenous glucose challenge was in-cluded and not additional insulin infusion to alter theplasma insulin profile. Buchanan et al3 have reportedthat the use of the minimal model without the additionof an intravenous insulin bolus was not able to detect in-sulin-mediated changes in the fractional glucose disap-pearance rate in 38% of women with gestational diabetesmellitus and 44% of pregnant women with normal glu-cose tolerance. Hence, when the minimal model isquoted as being a reasonable estimate of insulin sensitiv-ity in pregnancy, investigators have used a minimal

model that incorporates an intravenous insulin infusion20 minutes after the glucose injection. Even with this ad-dition of insulin to the minimal model, approximately10% of women will still not have a detectable insulin sen-sitivity value.3

Finally, we have previously reported data on 24-hoururinary insulin profiles in a subset of the women de-scribed in our article.4 In 383 women, pregnancy-associated hypertension without proteinuria developedin 67 (18%) and preeclampsia in 34 (9%). In the 67women in whom transient hypertension developed, the24-hour urine C-peptide level was significantly greater (P=.04) in midpregnancy, as compared with a controlgroup. In contrast, women with development ofpreeclampsia had decreased 24-hour urine C-peptide val-ues in late pregnancy as compared with those in the con-trol group (P =.03). We interpret these data as showingthat in women with pregnancy-associated hypertensionthere is increased insulin resistance in midpregnancycompared with the control group. Furthermore, we be-lieve that the significantly decreased 24-hour urine C-peptide level in women with development of preeclamp-sia in late pregnancy represents pancreatic β-celldysfunction, which is found in subjects with severe in-sulin resistance (for example, women with gestational di-abetes).3, 5 Further analysis of insulin in 24-hour urinespecimens is under way to determine whether similar pat-terns emerge in a larger study group.

We thank Caruso et al for their interest in our articleand look forward to their forthcoming publication.

Patrick M. Catalano, MDDepartment of Reproductive Biology, Case Western Reserve University atMetrohealth Medical Center, 2500 Metro Health Dr, Cleveland, OH44109

Gary M. Joffe, MDDepartment of Obstetrics and Gynecology, Lovelace Medical Center,5400 Gibson, SE, Albuquerque, NM 87108

Richard J. Levine, MDDivision of Epidemiology, Statistics, and Prevention Research, NationalInstitute of Child Health and Human Development, National Institutesof Health, 6100 Executive Blvd, Rockville, MD 20852

REFERENCES

1. Solomon CG, Graves SW, Greene MF, Seely EW. Glucose intol-erance as a predictor of hypertension in pregnancy.Hypertension 1994;23:717-21.

2. Roberts RN, Henrisken JE, Hadden DR. Insulin sensitivity inpreeclampsia. Br J Obstet Gynaecol 1998;105:1095-100.

3. Buchanan TA. Measurements of insulin sensitivity in pregnancywith glucose clamps and minimal model. In: Bergman RN,Lovejoy JC, editors. The minimal model approach to determi-nants of glucose to leucine. Baton Rouge (LA): Louisiana StateUniversity Press; 1997. p. 323-43.

4. Catalano PM, Petrulis A, Barabcach L, Amini S. Insulin resis-tance in nulliparous women developing preeclampsia [abstract262]. J Soc Gynecol Invest 1997;4:144A.

5. Catalano PM, Tyzbir ED, Wolf RR, Calles JE, Roman NM, AminiSB, et al. Carbohydrate metabolism in pregnancy in control sub-jects and women with gestational diabetes. Am J Physiol 1993;264(Endocrinol Metab 27):E60-7.

6/8/100143

Volume 181, Number 3 Letters 769Am J Obstet Gynecol