LETTERS

were 14, 82, and 113 pmol/l (normal ●●●●●●●●●●●●●●●●●●●●●●● OBSERVATIONS range 11–325), ruling out a condition of References iron overload. 1. Escobar-Morreale HF, Luque-Ramı´rez M, Forty-eight of the patients (61.5%) Alvarez-Blasco F, Botella-Carretero JI, and 24 of the control subjects (55.8%) Sancho J, San Milla´n JL: Body iron stores are increased in overweight and obese Mutations in the had one or more mutated alleles of the Hereditary women with polycystic ovary syndrome C282Y and H63D genotype alleles, (Brief Report). Care 28:2042– Hemochromatosis whereas all other women were homozy- 2044, 2005 gous for wild-type alleles of both HFE 2. Sanchez M, Bruguera M, Bosch J, Rodes J, Gene Are Not 2 mutations (␹ ϭ 0.377, P ϭ 0.539). The Ballesta F, Oliva R: Prevalence of the Associated With the HFE mutations studied here did not influ- Cys282Tyr and His63Asp HFE gene mu- Increased Body Iron ence serum ferritin levels when consider- tations in Spanish patients with heredi- ing PCOS patients and control subjects as tary hemochromatosis and in controls. Stores Observed in a whole (C282C [n ϭ 110] 109 Ϯ 94 J Hepatol 29:725–728, 1998 Overweight and pmol/l vs. C282Y [n ϭ 11]110 Ϯ 115 Obese Women With pmol/l [F ϭ 0.122, P ϭ 0.728]; H63H [n ϭ 57] 108 Ϯ 98 pmol/l vs. H63D and Polycystic Ovary D63D [n ϭ 64] 110 Ϯ 94 pmol/l [F ϭ Syndrome 0.499, P ϭ 0.481]; and interaction be- On the Weighted- tween both genotypes [Fϭ 0.834, P ϭ 0.363]) or separately (data not shown). Average e recently reported (1) that se- Finally, a multivariate stepwise linear Relationship rum ferritin levels are increased regression analysis model retained BMI Between Plasma W in overweight and obese women (␤ϭ0.263, P ϭ 0.003) and PCOS status with polycystic ovary syndrome (PCOS) (␤ϭ0.238, P ϭ 0.007) as predictive vari- Glucose and HbA1c independently of inflammation. This ables of serum ferritin levels (R2 ϭ 0.127, finding suggested increased body iron F ϭ 8.557, P Ͻ 0.001), whereas carrier stores in these women, raising the possi- status for C282Y and/or H63D mutations, bA1c (A1C) is widely used to assess bility that genes related to iron metabo- as well as having oligo/amenorrhea com- glycemic control in clinical and re- lism are altered in PCOS. pared with having regular cycles, were ex- H search settings, but the precise rela- Classic hereditary hemochromatosis cluded as predictors. tionship between A1C and preceding self- is an autosomal recessive disorder caused In summary, PCOS is not associated monitored plasma glucose measurements by mutations in the HFE gene, resulting in with the C282Y and H63D mutations in is recognized to be complex. It has been increased intestinal iron absorption and HFE, and these mutations did not influ- reported that measuring plasma glucose iron accumulation in several organs. In ence serum ferritin levels in our series. As levels in the 120 days before an A1C mea- the study by Sanchez et al. (2), Ͼ80% of discussed earlier (1), other mechanisms surement has a nonuniform effect on the the Spanish patients with hereditary are possibly related to the increase in result depending on the time that has hemochromatosis were homozygous for body iron stores observed in overweight elapsed between the glucose level and the HFE C282Y mutation or compound and obese PCOS patients. subsequent A1C measurement (1). heterozygotes for the HFE C282Y and Tahara and Shima (2) attempted to model 1 this weighted-average relationship between H63D mutations. JOSE´ I. BOTELLA-CARRETERO, MD, PHD 1 plasma glucose and A1C by measuring de- Although hereditary hemochromato- MANUEL LUQUE-RAMı´REZ, MD 1 creases in glucose and corresponding sis has low penetrance in young women, FRANCISCO A´ LVAREZ-BLASCO, MD 2 decreases in A1C in patients admitted to we studied the HFE genotypes of 78 JOSE´ L. SAN MILLAN´ , PHD 1 the hospital. Their model gives maximum PCOS patients and 43 control subjects HECTOR´ F. ESCOBAR-MORREALE, MD, PHD characterized in our previous report of in- weighting to glucose measurements im- 1 mediately before the A1C measurement, creased body iron stores in PCOS (1). From the Department of Endocrinology, Hospital Genotyping was conducted by PCR/ Ramo´n y Cajal, Madrid, Spain; and the 2Department with the weighting linearly decreasing for restriction fragment–length polymor- of Molecular Genetics, Hospital Ramo´n y Cajal, Ma- glucose measurements further back in phism methods using the PmlI and BclI drid, Spain. time, reaching zero weighting for plasma Address correspondence to He´ctor F. Escobar- glucose Ͼ120 days before the A1C. restriction enzymes for the C282Y and Morreale, Department of Endocrinology, Hospital H63D mutations, respectively. The ethics Ramo´n y Cajal, Carretera de Colmenar km 9Ј1, Trevin˜ o (3) reported that this weighted- committee of the Hospital Ramo´n y Cajal Madrid E-28034, Spain. E-mail: hescobarm.hrc@ average relationship leads to an anoma- approved the study, and informed con- salud.madrid.org. lous relationship between the exponential sent was obtained from all participants. DOI: 10.2337/dc06-1655 decay rates of glucose (G ) and A1C. We © 2006 by the American Diabetes Association. t We did not find homozygosity for the have reviewed this result and believe that C282Y substitution in HFE in any PCOS no such anomaly exists. Trevin˜ o sub- patient or control subject. Three patients Acknowledgments— This study was sup- tracted A1C calculated from the Tahara with PCOS but no control subjects were ported by grants FIS PI050341, PI050551, model (Ht) from “the mean of patient- compound heterozygotes for the C282Y and RGDM G03/212 from the Fondo de Inves- admission A1C values” (Hstart), obtaining 2 and H63D mutations (␹ ϭ 1.696, P ϭ tigacio´n Sanitaria, Instituto de Salud Carlos the counterintuitive result that a faster de- 0.552), but their serum ferritin levels III, Spain. cay in blood glucose results in a slower

2556 DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 Letters

Ϫ decay in (Hstart Ht). However, this “in- Is Pregnancy with and 532 consecutive verted” decay is likely to be due to sub- Outcome Worse in singleton pregnancies in women with tracting H from a constant value. His referred to the Diabetes t Type 2 Than in Type expression for Ht is an absolute A1C and Pregnancy Unit at University Hospi- value, not a change in A1C. An initial 1 Diabetic Women? tal La Paz from 1984 to 2004. value, Gs, has been specified for Gt, and Women with type 2 diabetes were hence an initial value is implicit in his significantly older ([means Ϯ SD] 31.8 Ϯ calculations. Subtracting H from a con- ost research on pregestational di- 5.5 vs. 29.4 Ϯ 4.7 years, P Ͻ 0.001), were t Ͻ stant would not be expected to give a valid abetes has focused on type 1 dia- more frequently obese (45.2 vs. 9%, P A1C estimate. betes, and surprisingly little 0.001), and had a shorter duration of di- M Ϯ Ϯ Ͻ To verify this conclusion, we simu- knowledge exists concerning outcomes of abetes (5.7 6 vs. 11.8 7.1 years, P pregnancies of women with type 2 diabe- 0.001). The rate of preconceptional care lated a patient with a constant glucose ϭ level followed by an exponential decay tes. A dearth of published data suggest (16.1 vs. 22.6%, P 0.175) and gesta- outcomes similar to those of type 1 dia- tional age at first visit (12.1 Ϯ 6.8 vs. upon admission to the hospital. During Ϯ ϭ the preadmission time period, the simu- betic women (1,2), although recent stud- 11.5 6.9 weeks’ gestation, P 0.529) did not differ between type 2 and type 1 lated A1C reached a steady state under the ies report poorer outcomes in women with type 2 diabetes (3–7). diabetic women. Maternal and perinatal constant glucose conditions, which We retrospectively compared mater- outcomes are shown in Table 1. avoided any ambiguity over the initial nal and perinatal outcomes of 93 consec- requirements and HbA (A1C) were value (H ) of A1C. In this simulation, 1c start utive singleton pregnancies in women lower during all three trimesters of preg- the decay rates of Ht then varied in the same way as those for the glucose data, as would be intuitively expected. The use of Table 1—Maternal and perinatal outcomes two initial values by Trevin˜ o, one for H and one for G, appears to have led to the anomalous result previously reported, Type 2 Type 1 rather than any inherent defect in the diabetes diabetes P weighted-average relationship proposed n* 93 532 by Tahara and Shima. Prepregnancy BMI (kg/m2) 28.9 Ϯ 6.5 23.3 Ϯ 3.1 Ͻ0.001 Maternal weight gain during 11.7 Ϯ 5.0 13.7 Ϯ 4.2 Ͻ0.001 pregnancy (kg) OLIVER J. GIBSON, MENG Glycemic control during pregnancy PATRICK E. MCSHARRY, DPHIL A1C at admission (%) 6.4 Ϯ 1.19 7.2 Ϯ 1.19 Ͻ0.001† LIONEL TARASSENKO, DPHIL A1C second trimester (%) 5.8 Ϯ 0.84 6.3 Ϯ 0.9 Ͻ0.001† A1C third trimester (%) 5.8 Ϯ 0.76 6.2 Ϯ 0.8 0.001† Insulin requirements From the Department of Engineering Science, Uni- Ϯ Ϯ Ͻ versity of Oxford, Oxford, U.K. First trimester (units/kg) 0.38 0.19 0.68 0.18 0.001† Address correspondence to Oliver J. Gibson, De- Second trimester (units/kg) 0.48 Ϯ 0.23 0.76 Ϯ 0.21 Ͻ0.001† partment of Engineering Science, University of Ox- Third trimester (units/kg) 0.62 Ϯ 0.31 0.93 Ϯ 0.26 Ͻ0.001† ford, Parks Road, Oxford, OX1 3PJ, U.K. E-mail: Pregnancy-induced hypertension 18 (19.4) 82 (15.4) 0.358 [email protected]. Preeclampsia 6 (6.5) 17 (3.2) 0.134 DOI: 10.2337/dc06-1646 Caesarean delivery 41 (44.1) 298 (56) 0.032 © 2006 by the American Diabetes Association. Gestational age (weeks of 37.1 Ϯ 1.6 36.7 Ϯ 1.7 0.018 gestation) ●●●●●●●●●●●●●●●●●●●●●●● Preterm delivery 28 (30.4) 186 (35.4) 0.406 Birth weight (g) 3,182 Ϯ 623 3,243 Ϯ 606 0.375 References Birth weight ratio 1.09 Ϯ 0.2 1.13 Ϯ 0.18 0.019 1. Rohlfing CL, Wiedmeyer H-M, Little RR, Large for gestational age 22 (23.9) 187 (35.6) 0.032 England JD, Tennill A, Goldstein DE: De- fining the relationship between plasma Small for gestational age 3 (3.3) 5 (1) 0.100 Perinatal mortality 1 (1.1) 9 (1.7) 1.000 glucose and HbA1c: analysis of glucose Major congenital malformations 6 (6.5) 25 (4.7) 0.442 profiles and HbA1c in the Diabetes Con- trol and Complications Trial. Diabetes Neonatal 24 (26.1) 172 (32.8) 0.226 Care 25:275–278, 2002 Neonatal hyperbilirrubinemia 34 (37) 223 (42.5) 0.359

2. Tahara Y, Shima K: Kinetics of HbA1c, gly- Neonatal hypocalcemia 4 (4.3) 30 (5.7) 0.805 cated albumin, and and Birth trauma 7 (7.6) 29 (5.5) 0.467 analysis of their weight functions against Neonatal sepsis 9 (9.8) 55 (10.5) 1.000 preceding plasma glucose level. Diabetes Neonatal polycythemia 14 (15.2) 71 (13.5) 0.626 Care 18:440–447, 1995 Neonatal respiratory distress 7 (7.6) 95 (18.1) 0.010 3. Trevin˜ o G: On the weighted-average rela- syndrome tionship between plasma glucose and Data are means Ϯ SD or n (%) unless otherwise indicated. *In the case of stillbirths (one in type 2 and seven HbA1c (Letter). Diabetes Care 29:466, in type 1 diabetes), no perinatal data other than the presence of major congenital malformations was 2006 analyzed; †adjusted for multiplicity.

DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 2557 Letters nancy in type 2 diabetic women. Maternal PILAR M. VAQUERO, MD Glucose weight gain and the rate of caesarean de- AFRICA VILLARROEL, MD Abnormalities in liveries were lower in type 2 diabetes. ALBERTO FERNANDEZ, BM Gestational age at birth was significantly LUIS F. PALLARDO, MD, PHD Patients With higher and the rate of large infants for ges- Hepatitis C Virus tational age lower in infants of women From the Division of Diabetes, Department of En- : with type 2 diabetes. The rates of perinatal docrinology and Nutrition, University Hospital La mortality and major congenital malfor- Paz, Madrid, Spain. Epidemiology and mations were comparable in both groups. Address correspondence to Natalia Hillman Pathogenesis First-trimester A1C in type 2 and type 1 Gadea, Division of Diabetes, Department of Endo- diabetic mothers with perinatal mortality crinology and Nutrition, University Hospital La Paz, Ϯ 28046 Madrid, Spain. E-mail: nhillman.hulp@ was 9.9 and 8.1 1.2%, respectively. salud.madrid.org. Response to Lecube et al. Among pregnancies complicated by ma- DOI: 10.2337/dc06-0680 jor congenital malformations, first- © 2006 by the American Diabetes Association. Ͼ trimester A1C was 7% in 84% of e read with great interest the re- women with type 1 diabetes and only in view article by Lecube et al. (1) one woman (16.7%) with type 2 diabetes ●●●●●●●●●●●●●●●●●●●●●●● on the pathogenic factors specif- ϭ W (P 0.006). Neonatal distress respiratory References ically linking hepatitis C virus (HCV) in- syndrome was more frequent in infants of 1. Feig DS, Palda VA: Type 2 diabetes in fection and glucose abnormalities. After mothers with type 1 diabetes. pregnancy: a growing concern. Lancet analyzing the different mechanisms by In our study, pregnancy outcomes in 359:1690–1692, 2002 which HCV is thought to contribute to the type 2 diabetic women were, if anything, 2. Group, France: development of type 2 diabetes, Lecube et similar to those with type 1 diabetes. In French multicentric survey of outcome of al. focus their attention on the role of fact, women with type 2 diabetes had pregnancy in women with pregestational proinflammatory , in particular diabetes. Diabetes Care 26:2990–2993, lower rates of large infants for gestational tumor necrosis factor (TNF)-␣ and inter- 2003 age, neonatal respiratory distress syn- leukin-6. They suggest that the activation drome, and caesarean delivery. 3. Omori Y, Minei S, Testuo T, Nemoto K, ␣ Shimizu M, Sanaka M: Current status of of the TNF- system in HCV-infected pa- As in some of the studies available, we tients, which has been directly related to found no significant differences in perina- pregnancy in diabetic women: a compar- in their recent study (2), tal mortality or major congenital malfor- ison of pregnancy in IDDM and NIDDM could be related to the T-helper (Th)1 im- mations between women with type 2 and mothers. Diabetes Res Clin Pract 24 mune response observed in the course of type 1 diabetes (1–2). However, the re- (Suppl.):S273–S278, 1994 HCV infection. Accordingly, as shown in sults of five recent publications (3–7) sug- 4. Brydon P, Smith T, Proffitt M, Gee H, Holder R, Dunne F: Pregnancy outcome Fig. 1 of their review article, the activation gest that type 2 diabetes could even in women with type 2 diabetes mellitus of the TNF-␣ system following the Th1 represent a higher risk of perinatal mor- needs to be addressed. Int J Clin Pract 54: tality or congenital malformations than immune-mediated response is central to 418–419, 2000 the pathogenesis of both liver fibrosis and that conferred by type 1 diabetes. Similar 5. Cundy T, Gamble G, Townend K, Henley rates of preconceptional care in women insulin resistance associated with HCV PG, MacPherson P, Roberts AB: Perinatal infection. with type 1 and type 2 diabetes in our mortality in type 2 diabetes mellitus. Dia- study could explain this discrepancy, as However, an apparent paradox is bet Med 17:33–39, 2000 raised by an attempt to fit such interpre- could the fact that gestational age at first 6. Clausen TD, Mathiesen E, Ekbom P, Hell- visit to the clinic was comparable in both muth E, Mandrup-Poulsen T, Damm P: tation with well-acquired data and the type 1 and type 2 diabetic women who Poor pregnancy outcome in women with most recent evidence from literature. In- did not undergo preconceptional care. type 2 diabetes. Diabetes Care 28:323– deed, a vigorous Th1 response In our study, congenital malforma- 328, 2005 has been classically observed in patients tions in type 2 diabetes were not related to 7. Roland JM, Murphy HR, Ball V, North- who clear their HCV infection, either poor first-trimester metabolic control in cote-Wright J, Temple RC: The pregnan- spontaneously (3) or in response to anti- most cases. The concurrence in women cies of women with type 2 diabetes: poor viral treatment (4,5). By contrast, recent with type 2 diabetes of factors other than outcomes but opportunities for improve- studies have demonstrated that insulin re- glycemic control, such as obesity and ment. Diabet Med 22:1774–1777, 2005 sistance is independently associated with older age, may account for this finding 8. Garcia-Patterson A, Erdozain L, Ginovart a poor response to antiviral therapy in (8). G, Adelantado JM, Cubero JM, Gallo G, de HCV patients (6,7), consistent with pre- In conclusion, our study shows that Leiva A, Corcoy R: In human gestational vious observations on the lower success pregnancy outcomes in type 2 diabetes diabetes mellitus congenital malforma- rate of interferon alone or interferon plus tions are related to pre-pregnancy body are better than in type 1 diabetes when ribavirin in obese and diabetic patients. mass index and to severity of diabetes. Therefore, it is difficult to understand type 2 diabetic women receive as much Diabetologia 47:509–514, 2004 intensified medical treatment during pre- how an increased Th1 immune response, conception and pregnancy as that given to which is protective in relation to viral type 1 diabetic women. clearance, can be, at the same time, the COMMENTS AND major determinant of insulin resistance NATALIA HILLMAN, MD and responsible for a poor response to an- LUCRECIA HERRANZ, MD, PHD RESPONSES tiviral treatment.

2558 DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 Letters

UMBERTO VESPASIANI GENTILUCCI, MD Glucose there is a cluster of alterations associated ANTONIO PICARDI, PHD Abnormalities in with insulin resistance such as obesity, PAOLO POZZILLI, MD ageing, hypertriglyceridemia, liver es- Patients With teatosis, and fibrosis; these alterations are Hepatitis C Virus also risk factors for nonresponse to anti- From the Department of Internal Medicine and viral treatment. It has recently been dem- Hepatology, University Campus Bio-Medico, Rome, Infection: onstrated (11) that hyperinsulinemia Italy. Address correspondence to Umberto Vespasiani Epidemiology and blocks the inhibition of HCV virus repli- Gentilucci, MD, University Campus Bio-Medico, Pathogenesis cation by interferon. Therefore, it seems Via E. Longoni, 83, I-00155 Rome, Italy. E-mail: that there is a vicious circle in which in- [email protected]. sulin resistance facilitates the persistence DOI: 10.2337/dc06-1374 Response to Gentilucci et al. of HCV and, alternatively, HCV favors in- © 2006 by the American Diabetes Association. sulin resistance. Altogether, one can depict a complex e thank Gentilucci et al. (1) for scenario in which Th1 response is only their comments on our articles one more of the actors. Future studies are ●●●●●●●●●●●●●●●●●●●●●●● W (2,3) regarding the pathogenic needed to not only confirm that insulin References mechanisms of diabetes in patients with resistance and type 2 diabetes are poor 1. Lecube A, Hernandez C, Genesca J, Simo hepatitis C virus (HCV) infection. The au- response predictors of antiviral treatment R: Glucose abnormalities in patients with thors question why an increased T-helper but also to unravel the mechanisms hepatitis C virus infection: epidemiology (Th)1 immune response can be simulta- involved. and pathogenesis. Diabetes Care 29: neously the major determinant of insulin 1140–1149, 2006 resistance and responsible for a poor re- RAFAEL SIMO´, MD 2. Lecube A, Hernandez C, Genesca J, sponse to antiviral treatment. This ques- Simo R: Proinflammatory cytokines, in- tion is based on the statement that Th1 From the Division of Endocrinology, Diabetes Re- sulin resistance, and insulin secretion in immunoresponse favors HCV clearance. search Unit, Hospital Universitari Vall d’Hebron, Universitat Auto`noma de Barcelona, Barcelona, chronic hepatitis C patients: a case-con- However, although a vigorous Th1 re- Spain. trol study. Diabetes Care 29:1096– sponse could play an essential role in Address correspondence to Dr. Rafael Simo´, Di- 1101, 2006 spontaneous viral clearance, this is not so abetes Research Unit, Endocrinology Division, Hos- 3. Pape GR, Gerlach TJ, Diepolder HM, evident after interferon treatment. It pital Vall d’Hebron, Pg. Vall d’Hebron 119-129, Gruner N, Jung M, Santantonio T: Role of should be noted that in sustained re- 08035 Barcelona, Spain. E-mail: [email protected]. the specific T-cell response for clearance net. sponders, pretreatment intrahepatic DOI: 10.2337/dc06-1563 and control of hepatitis C virus. J Viral mRNA levels of ␥-interferon and tumor © 2006 by the American Diabetes Association. Hepat 6 (Suppl. 1):36–40, 1999 necrosis factor-␣ were lower than in non- 4. Sarobe P, Jauregui JI, Lasarte JJ, Garcia N, sustained responders (4). In addition, a ●●●●●●●●●●●●●●●●●●●●●●● Civeira MP, Borras-Cuesta F, Prieto J: References Production of interleukin-2 in response lower Th2 response during antiviral treat- ment (specifically a decrease in interleu- 1. Gentilucci UV, Picardi A, Pozzilli P: Glu- to synthetic peptides from hepatitis vi- cose abnormalities in patients with hepa- rus E1 protein in patients with chronic kin [IL]-10 rather than an increase of Th1) has been associated with a long-term titis C virus infection: epidemiology and hepatitis C: relationship with the re- pathogenesis (Letter). Diabetes Care sponse to interferon treatment. J Hepatol virological response (5,6). Tsai et al. (7) 29:2558–2559, 2006 25:1–9, 1996 and Eckels et al. (8) demonstrated that in 2. Lecube A, Herna´ndez C, Genesca` J, Simo´ 5. Cramp ME, Rossol S, Chokshi S, Carucci vitro cytokine responses to recombinant R: Glucose abnormalities in patients with P, Williams R, Naoumov NV: Hepatitis C HCV antigens were confined to IL-4 and hepatitis C virus infection: epidemiology virus-specific T-cell reactivity during in- IL-10 and proposed that such Th2 pre- andpathogenesis.DiabetesCare29:1140– terferon and ribavirin treatment in dominance might be conductive to viral 1149, 2006 chronic hepatitis C. Gastroenterology 118: persistence. Furthermore, Masaki et al. 3. Lecube A, Herna´ndez C, Genesca` J, Simo´ 346–355, 2000 (9) reported that a lower Th1/Th2 ratio R: Proinflammatory cytokines, insulin re- 6. Romero-Gomez M, Del Mar Viloria M, before interferon therapy may favor long- sistance and insulin secretion in chronic hepatitis C patients: a case-control study. Andrade RJ, Salmeron J, Diago M, Fer- term virological response in patients with nandez-Rodriguez CM, Corpas R, Cruz Diabetes Care 29:1096–1101, 2006 chronic hepatitis C. In addition, activa- 4. Dumoulin FL, Wennrich U, Nischalke M, Grande L, Vazquez L, Munoz-De- tion of naı¨ve B-cells via CD81 has been Rueda P, Lopez-Serrano P, Gila A, Gutier- HD, Leifeld L, Fisher HP, Sauerbruch T, involved in the immunological response Spengler U: Intrahepatic mRNA levels of rez ML, Perez C, Ruiz-Extremera A, triggered by HCV (10). Therefore, the interferon gamma and tumor necrosis fac- Suarez E, Castillo J: Insulin resistance im- tor alpha and response to antiviral treat- pairs sustained response rate to peginter- immune mechanisms involved in the clearance of HCV after interferon therapy ment of chronic hepatitis C. J Hum Virol feron plus ribavirin in chronic hepatitis C 4:195–199, 2001 patients. Gastroenterology 128:636–641, are complex and are far from being elucidated. 5. Cramp ME, Rossol S, Chokshi S, Carucci 2005 P, Williams R, Naoumov NV: Hepatitis C 7. D’Souza R, Sabin CA, Foster GR: Insulin Low-grade inflammation mediated by virus-specific T-cell reactivity during in- resistance plays a significant role in liver activated innate immunity is an underly- terferon and ribavirin treatment in fibrosis in chronic hepatitis C and in the ing pathogenic mechanism of insulin re- chronic hepatitis C. Gastroenterology 118: response to antiviral therapy. Am J Gastro- sistance and type 2 diabetes. Apart from 346–355, 2000 enterol 100:1509–1515, 2005 the impairment of immune response, 6. Torre F, Rossol S, Pelli N, Basso M,

DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 2559 Letters

Delfino A, Picciotto A: Kinetics of soluble ment of clinically diagnosed type 2 School of Medicine and Pharmacology, University of tumour necrosis factor (TNF)-alpha re- diabetes. The authors state that LADA pa- Western Australia, Fremantle Hospital, P.O. Box ceptors and cytokines in the early phase of 480, Fremantle, Western Australia, 6959, Australia. tients can require rapid escalation of oral E-mail: [email protected]. treatment for chronic hepatitis C: com- therapy or early commencement of insu- parison between interferon (IFN)-alpha DOI: 10.2337/dc06-1321 lin (1). However, patients with severely © 2006 by the American Diabetes Association. alone, IFN-alpha plus amantadine or plus deficient ␤-cell function but insufficient ribavirine. Clin Exp Immunol 136:507– ●●●●●●●●●●●●●●●●●●●●●●● LADA features still need insulin therapy. 512, 2004 References 7. Tsai SL, Liaw YF, Chen MH, Huang CY, In addition, some LADA patients achieve reasonable initial glycemic control with 1. Fourlanos S, Perry C, Stein MS, Stankov- Kuo GC: Detection of type 2-like T helper ich J, Harrison LC, Colman PG: A clinical cells in hepatitis C virus infection: impli- oral agents (2), with insulin available screening tool identifies autoimmune di- cations for hepatitis C virus chronicity. should this strategy fail. abetes in adults. Diabetes Care 29:970– Hepatology 25:449–458, 1997 We have concerns that the LADA in- 975, 2006 8. Eckels DD, Tabatabail N, Bian TH, Wang strument fails to meet the necessary crite- 2. Davis TM, Wright AD, Mehta ZM, Cull H, Muheisen SS, Rice CM, Yoshizawa K, ria for a valid screening tool (3). In their CA, Stratton IM, Bottazzo GF, Bosi E, Gill J: In vitro human Th-cell responses to small study, Fourlanos et al. report a sen- Mackay IR, Holman RR: Islet autoanti- a recombinant hepatitis C virus antigen: sitivity of 90% and specificity of 71%. bodies in clinically diagnosed type 2 dia- failure in IL-2 production despite prolif- However, the positive predictive value is betes: prevalence and relationship with eration. Hum Immunol 60:187–199, 1999 metabolic control (UKPDS 70). Diabetolo- 9. Masaki N, Fukushima S, Hayashi S: 21%, indicating that the probability of correctly diagnosing LADA is low. This, gia 48:695–702, 2005 Lower th1/th2 ratio before interferon 3. Barratt A, Irwig L, Glasziou P, Cumming therapy may favor long-term virological and the high false-positive rate (28%), RG, Raffle A, Hicks N, Gray JAM, Guyatt responses in patients with chronic hepa- suggest a limited ability to identify pa- GH: Users’ guide to the medical literature. titis C. Dig Dis Sci 47:2163–2169, 2002 tients most in need of early insulin therapy. XVII. How to use guidelines and recom- 10. Rosa D, Saletti G, De Gregorio E, Zorat F, The authors’ apparent intention is to mendations about screening. JAMA 281: Comar C, D’Oro U, Nuti S, Houghton M, promote the instrument as part of usual 2029–2034, 1999 Barnaba V, Pozzato G, Abrignani S: Acti- care. Because of this, and since the Amer- 4. American Diabetes Association: Stan- vation of naı¨ve B lymphocytes via CD81, a ican Diabetes Association does not rec- dards of medical care in diabetes (Position pathogenic mechanism for hepatitis C vi- ommend islet antibody testing in type 2 Statement). Diabetes Care 29 (Suppl. 1): rus-associated B lymphocyte disorders. S4–S42, 2006 Proc Natl Acad Sci U S A 102:18544– diabetes (4), why do the authors recom- 18549, 2005 mend serological confirmation (1)? Even 11. Sanyal AJ, Chand N, Comar K, Mirshahi in the case of children, in whom educa- A Clinical Screening F: Hyperinsulinemia blocks the inhibition tion, dietary counsel, and treatment differ of hepatitis C virus (HCV) replication by markedly by diabetes type, autoantigens Tool Identifies interferon: a potential mechanism for fail- may be present in a substantial number Autoimmune ure of interferon therapy in subjects with with otherwise straightforward type 2 di- Diabetes in Adults HCV and nonalcoholic fatty liver disease abetes (4). One reason for antibody test- (Abstract). Hepatology 40 (Suppl. 1):179A, ing may be to characterize LADA patients 2004 fully for intervention studies (1), but this Response to Davis et al. would only be appropriate in specialist A Clinical Screening centers. e thank Davis et al. (1) for their We contend that the management of comments regarding the recent Tool Identifies poorly controlled type 2 diabetes in adults W publication of a clinical screen- Autoimmune should be based on detailed clinical assess- ing tool for latent autoimmune diabetes in Diabetes in Adults ment (including the LADA instrument adults (LADA) (2). It is appreciated that components), review of glycemic control, the authors’ routine management of implementation of strategies (including ed- “poorly controlled adult type 2 diabetes” Response to Fourlanos et al. ucator and dietitian input) that might im- incorporates the “LADA instrument com- prove adherence to self-management, a ponents.” However, our observations of ourlanos et al. (1) report on a screen- discussion of available therapies (includ- the management of such patients by inter- ing instrument designed to facilitate ing insulin), and adequate monitoring nists and diabetes nurse practitioners in F management of latent autoimmune and support. The use of the LADA instru- the community are often contrary to the diabetes of adults (LADA). They assert ment and/or autoantibody testing appears practice of the authors. Adults with sub- that in poorly controlled type 2 patients redundant in this setting. optimal glycemic control due to declining exhibiting two or more of five features ␤-cell function (often secondary to auto- Ͻ 1 (age 50 years, hyperglycemic symp- TIMOTHY M.E. DAVIS, FRACP immune disease) are underrecognized, Ͻ 2 2 toms, BMI 25.0 kg/m , and personal CAROLE A. CULL, PHD leading to delays in commencing insulin 2 and family history of autoimmunity), the RURY R. HOLMAN, FRACP therapy. The clinical screening tool was “logical ” next step is confirmatory islet developed to aid primary care physicians antibody testing (1). From the 1School of Medicine and Pharmacology, and diabetes nurse practitioners to con- Although the effect of routine use of University of Western Australia, Fremantle Hospi- sider the pathophysiological process of tal, Fremantle, Western Australia, Australia; and the ␤ the instrument on outcomes such as 2Diabetes Trial Unit, University of Oxford, Oxford, autoimmune -cell destruction. The au- HbA1c is unknown, the article raises im- Oxfordshire, U.K. thors cite that the positive predictive portant questions relating to the manage- Address correspondence to Prof. Tim Davis, value of the clinical screening tool is low

2560 DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 Letters at 21% but do not mention that the neg- (1). The composite primary end point rate ●●●●●●●●●●●●●●●●●●●●●●● ative predictive value of the tool is 99%; (10 mg/day atorvastatin versus placebo) References hence, the tool is highly reliable at exclud- showed a hazard ratio of 0.90 (95% CI 1. Knopp RH, D’Emden M, Smilde JG, Po- ing LADA and has a sensitivity of 90%, 0.73–1.12, P ϭ 0.34) after 4 years. Knopp cock SJ, the ASPEN Study Group: Efficacy meaning that most LADA patients can be et al. (1) highlight some of the differences and safety of atorvastatin in the preven- tion of cardiovascular end points in sub- identified with the assistance of this nonin- between ASPEN and previous atorvasta- jects with type 2 diabetes: the Atorvastatin vasive and cost-free clinical screening tool. tin trials (Collaborative Atorvastatin Dia- Study for Prevention of Coronary Heart 1,2 betes Study and Anglo-Scandinavian SPIROS FOURLANOS Disease Endpoints in Non-Insulin-De- 1 Cardiac Outcomes Trial) also involving pendent Diabetes Mellitus (ASPEN). Dia- LEONARD C. HARRISON 1,2 diabetic individuals without established betes Care 29:1478–1485, 2006 PETER G. COLMAN coronary heart disease (2,3). 2. Colhoun HM, Betteridge DJ, Durrington PN, Hitman GA, Neil HA, Livingstone SJ, From the1Autoimmunity and Transplantation Divi- Other differences may also be rele- sion, The Walter and Eliza Hall Institute of Medical vant. In ASPEN, 78.3% of those on ator- Thomason MJ, Mackness MI, Charlton- Research, Parkville, Victoria, Australia; and the 2De- vastatin and 76.4% of those in the placebo Menys V, Fuller JH: Primary prevention of partment of Diabetes and Endocrinology, The group were included in the analysis. This cardiovascular disease with atorvastatin Royal Melbourne Hospital, Parkville, Victoria, in type 2 diabetes in the Collaborative Australia. represents a substantial “drop-out” rate. Atorvastatin Diabetes Study (CARDS): Address correspondence to Spiros Fourlanos, Di- Furthermore, by the end of the study, multicentre randomised placebo-con- vision of Autoimmunity and Transplantation, medication was taken by 67.5% of those trolled trial. Lancet 364:685–696, 2004 Walter and Eliza Hall Institute of Medical Research, in the atorvastatin group and 57.6% of 3. Sever PS, Poulter NR, Dahlo¨f B, Wedel H, 1G Royal Parade Melbourne, Victoria VIC 3050, those in the placebo group. The “drop-in” Collins R, Beevers G, Caulfield M, Kjeld- Australia. E-mail: [email protected]. DOI: 10.2337/dc06-1565 rate in ASPEN was also high; 26.9% of sen SE, Kristinsson A, McInnes GT, © 2006 by the American Diabetes Association. those on placebo and 15.4% of those in Mehlsen J, Nieminen M, O’Brien E, Oster- ●●●●●●●●●●●●●●●●●●●●●●● the atorvastatin group took concomitant gren J, the ASCOT Investigators: Reduc- hypolipidemic agents. Nevertheless, LDL tion in cardiovascular events with References cholesterol was reduced by 29% with atorvastatin in 2,532 patients with type 2 diabetes: Anglo-Scandinavian Cardiac 1. Davis TME, Cull CA, Holman RR: A clin- atorvastatin relative to placebo. Is it pos- ical screening tool identifies autoimmune Outcomes Trial–Lipid-Lowering Arm sible that among the patients on atorva- (ASCOT-LLA). Diabetes Care 28:1151– diabetes in adults (Letter). Diabetes Care statin, some took a second statin? If so, 29:2560, 2006 1157, 2005 how many of the placebo-treated patients 2. Fourlanos S, Perry C, Stein MS, Stankov- were taking a statin and for how long? ich J, Harrison LC, Colman PG: A clinical screening tool identifies autoimmune di- In the ASPEN study (1), blood pres- Efficacy and Safety abetes in adults. Diabetes Care 29:970– sure was well controlled (mean 133/77 of Atorvastatin in 975, 2006 mmHg). The blood pressure in the Col- laborative Atorvastatin Diabetes Study the Prevention of and the Anglo-Scandinavian Cardiac Cardiovascular End Efficacy and Safety Outcomes Trial was ϳ138/78 and 143/80 of Atorvastatin in mmHg, respectively (2,3). This difference Points in Subjects the Prevention of may influence any benefit accruing from With Type 2 lipid lowering in ASPEN. There was also a Diabetes: The Cardiovascular End change in protocol during the ASPEN Points in Subjects study. Did this lead to a difference in the Atorvastatin Study With Type 2 duration of follow-up in the primary and for Prevention of secondary prevention groups? Coronary Heart Diabetes: The The differences outlined above, to- Atorvastatin Study gether with those mentioned by the Disease Endpoints in for Prevention of ASPEN authors (1), may have contributed Non-Insulin- to the nonsignificant reduction in events Dependent Diabetes Coronary Heart reported in this trial. Disease Endpoints in Mellitus (ASPEN) IRENE F. GAZI, MD Non-Insulin- IMITRI IKHAILIDIS FRCP D P. M , Response to Gazi and Mikhailidis Dependent Diabetes From the Department of Clinical Biochemistry, Mellitus (ASPEN) Royal Free Hospital, Royal Free and University Col- lege of Medicine, University of London, London, e appreciate the interest of Gazi U.K. and Mikhailidis (1) in the Ator- Response to Knopp Address correspondence to Dr. Dimitri P. Mikhailidis MD, FFPM, FRCP, FRCPath, Reader and W vastatin Study for Prevention of Honorary Consultant, Department of Clinical Bio- Coronary Heart Disease Endpoints in e read with interest the results of chemistry, Royal Free Hospital, Royal Free and Uni- Non-Insulin-Dependent Diabetes Melli- the Atorvastatin Study for Pre- versity College of Medicine, University of London, tus (ASPEN) and their proposed reasons vention of Coronary Heart Dis- Pond Street, London NW3 2QG, U.K. E-mail: for the nonsignificant results (2). W [email protected]. ease Endpoints in Non-Insulin- DOI: 10.2337/dc06-1487 We mention in our article the high Dependent Diabetes Mellitus (ASPEN) © 2006 by the American Diabetes Association. rates of treatment “drop in” and “drop

DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 2561 Letters out” and the potential impact on the pri- sults has many possible explanations. but limited by the patients examined, mary end point (mostly due to changing Nonetheless, the ASPEN study reminds who were predominantly male (98%), guidelines and study design changes). us that the many risk factors for heart dis- mainly with type 2 diabetes, and were re- While Gazi and Mikhailidis correctly state ease in diabetes require individualized cruited from a hospital diabetes clinic. We that there was a high incidence of drop- management for a complete treatment have already addressed these problems in outs, we should clarify that all subjects approach. a previous publication (2) using a similar, (including those who withdrew) were in- clinically focused foot prediction ROBERT H. KNOPP, MD cluded in the final analysis. The authors tool (3) that included many of the criteria ON BEHALF OF THE ASPEN STUDY GROUP ask if the substantial differential LDL cho- recommended by the International lesterol decrease of 29% between active From the University of Washington School of Med- Working Group on the (4). and placebo groups could reflect active icine, Seattle, Washington. Our grading scheme categorized 3,526 subjects taking a second statin. Of the Address correspondence to Robert H. Knopp, patients into low, moderate, or high risk MD, Chief, Division of Metabolism, Endocrinology 15% of atorvastatin-treated subjects tak- and Nutrition, Harborview Medical Center, 325 of ulceration. High-risk patients “were 83 ing concomitant lipid-lowering medica- Ninth Ave., 359720, Seattle, WA 98104-2499. E- times more likely to ulcerate than low tions, the vast majority took an additional mail: [email protected]. risk” patients, and the chance of “low- statin. However, considered on its own, R.H.K. has served on an advisory board for, has risk” patients remaining ulcer-free after received honoraria from, and has received grant/ this would not explain the 29% reduction research support from Pfizer. 2.4 years was 99.7% (2). This tool was in LDL cholesterol compared with pla- DOI: 10.2337/dc06-1678 valid for type 1 and type 2 diabetic male cebo. Of the 26.9% of subjects in the pla- © 2006 by the American Diabetes Association. and female subjects in a population-based cebo group who were taking concomitant cohort. Such foot ulcer prediction tools lipid-lowering medications (mostly ●●●●●●●●●●●●●●●●●●●●●●● are thus useful for “all-comers” in a gen- statins), 19.7% took them for Ն30 days. References eral community setting, as well as in spe- It is likely more important that 42% of 1. Gazi IF, Mikhailidis DP: Efficacy and cialized hospital clinics. subjects with cardiovascular events in the safety of atorvastatin the prevention of Boyko et al. also raised the issue that atorvastatin group had stopped their ran- cardiovascular end points in subjects with patients at high risk of ulceration may be domized medication Ͼ1 year before their type 2 diabetes: the Atorvastatin Study for at increased risk of death. We demon- Prevention of Coronary Heart Disease event. Endpoints in Non-Insulin-Dependent Di- strated that the crude mortality rate for We agree that lower blood pressure, abetes Mellitus (ASPEN) (Letter). Diabetes high-risk patients was 19.1% compared as well as lower baseline LDL cholesterol, Care 29:2561–2562, 2006 with 3.4% for low-risk patients (2). Thus, younger age, lower smoking rates, and a 2. Knopp RH, D’Emden M, Smilde JG, Po- high risk of ulceration is associated with smaller proportion of men combined to cock SJ, the ASPEN Study Group: Efficacy increased death rate as suspected by place primary prevention subjects in the and safety of atorvastatin the prevention Boyko et al., which may result in an un- ASPEN at lower CVD risk than those in of cardiovascular end points in subjects derestimation of the predictive value of the Collaborative Atorvastatin Diabetes with type 2 diabetes: the Atorvastatin these clinical tools, as patients may die Study (3). Despite this apparent lower Study for Prevention of Coronary Heart before they develop foot ulcers. risk, a greater incidence of cardiovascular Disease Endpoints in Non-Insulin-De- These two studies complement each pendent Diabetes Mellitus (ASPEN). Dia- events was observed in placebo-treated betes Care 29:1478–1485, 2006 other by demonstrating that the overall primary prevention subjects in ASPEN 3. Colhoun HM, Betteridge DJ, Durrington foot ulcer risk assessment is greater than (10.8%) than in CARDS (9.0%), indicat- PN, Hitman GA, Neil HA, Livingstone SJ, any individual criteria (1) and that the ing inclusion of “softer” end points, such Thomason MJ, Mackness MI, Charlton- tool is valid in routine clinical practice for as hospitalization for angina pectoris and Menys V, Fuller JH: Primary prevention of all patients in the community (2) and spe- interventions. Nonetheless, trends in cardiovascular disease with atorvastation cialized centers (1,2). Foot ulcer predic- CVD event reduction with atorvastatin in type 2 diabetes in the Collaborative tion tools may be useful in directing were at the expected rates for the fatal and Atorvastatin Diabetes Study (CARDS): educational initiatives and scarce health nonfatal myocardial infarction end point multicentre randomized placebo-con- care resources to those at greatest need. and in the secondary prevention cohort trolled trial. Lancet 364:685–696, 2004 1 (2). GRAHAM P. LEESE, MD 2 The subjects in the secondary preven- Prediction of ANDREW D. MORRIS, MD tion group entered the study before the Diabetic Foot Ulcer From the 1Ninewells Hospital and Medical School primary prevention group. Secondary Dundee, Dundee, Scotland; and the 2Department of prevention subjects would have remained Occurrence Using Medicine, University of Dundee, Dundee, Scotland. in the study longer were it not for the Commonly Available Address correspondence to Dr. Graham Leese, Safety and Data Monitoring Board recom- Ward 1 and 2, Ninewells Hospital and Medical mendation late in the study to stop the Clinical Information School, Dundee, Scotland, U.K. DD1 9SY. E-mail: [email protected]. study drug in the secondary prevention DOI: 10.2337/dc06-1471 cohort and begin active treatment. As a Response to Boyko et al. © 2006 by the American Diabetes Association. result, the durations of follow-up were similar: 4.50 and 4.38 years for secondary ●●●●●●●●●●●●●●●●●●●●●●● versus primary prevention subjects taking n their article, Boyko et al. (1) describe References atorvastatin and 4.38 and 4.46 years, re- a foot ulcer prediction tool that will be 1. Boyko EJ, Ahoroni JH, Cohen V, Nelson spectively, for those taking placebo. I useful in practice, as it is based on simple KM, Heagerty PJ: Prediction of diabetic The nonsignificance of the ASPEN re- clinical criteria. The tool is well validated foot ulcer occurrence using commonly

2562 DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 Letters

available clinical information. Diabetes clinics; they also suggest that their find- ●●●●●●●●●●●●●●●●●●●●●●● Care 29:1202–1207, 2006 ings are valid for male and female subjects References 2. Leese GP, Reid F, Green V, McAlpine R, and for both types of diabetes (1). How- 1. Leese GP, Morris A: Prediction of diabetic Cunningham S, Emslie-Smith A, Morris foot ulcer occurrence using commonly AD, McMurray B, Connacher A: Predict- ever, their data and analysis do not pro- available clinical information (Letter). Di- ing foot ulceration in diabetes: validation vide strong support for these statements abetes Care 29:2562–2563, 2006 of a clinical tool in a population-based for several reasons. First, although they 2. Leese GP, Reid F, Green V, McAlpine R, study. Int J Clin Prac 60:541–545, 2006 used a population-sampling strategy that Cunningham S, Emslie-Smith AM, Morris 3. Scottish Intercollegiate Guideline Net- targeted 8,923 subjects with diabetes, AD, McMurray B, Connacher AC: Strati- work (SIGN): SIGN Guideline Publication only 3,526 (40%) underwent a clinical fication of foot ulcer risk in patients with No.55: The Management of Diabetes. Edin- foot risk assessment. Thus, most subjects diabetes: a population-based study. Int burgh, U.K., Royal College of Physicians, J Clin Pract 60:541–545, 2006 Edinburgh, 2001. Available from http:// in the sampling frame were not included 3. Boyko EJ, Ahroni JH, Cohen V, Nelson http://www.sign.ac.uk/guidelines/ in the analysis, which certainly raises con- KM, Heagerty PJ: Prediction of diabetic fulltext/55/index.html. Accessed 19 cerns regarding the validity of findings foot ulcer occurrence using commonly September 2006 and limits the degree of confidence with available clinical information: the Seattle 4. International Diabetes Federation: Inter- which one can recommend these results Diabetic Foot Study. Diabetes Care national Concensus on the Diabetic Foot: for “all-comers.” Second, although both 29:1202–1207, 2006 Five Cornerstones, Identification of the Foot sexes and diabetes types were included in 4. Boyko EJ, Ahroni JH, Stensel V, Forsberg at Risk (Consensus Meeting). May 2003. RC, Davignon DR, Smith DG: A prospec- Available from http://www.iwgdf.org/ their study, the appropriate analysis of in- tive study of risk factors for diabetic foot index.php?optionϭcom_content&task- teraction to determine whether the results ulcer: the Seattle Diabetic Foot Study. Di- view&idϭ39&Itemidϭ60. Accessed 19 apply similarly in these subgroups of in- abetes Care 22:1036–1042, 1999 September 2006 terest was not performed. 5. Boyko EJ, Ahroni JH, Smith DG, Dav- Leese and Morris bring up the issue of ignon D: Increased mortality associated competing risks, in that individuals at with diabetic foot ulcer. Diabet Med 13: 967–972, 1996 Prediction of higher risk for diabetic foot ulcer are also at higher risk for death. They refer to our Diabetic Foot Ulcer “suspicion” that the death rate in people at Occurrence Using high risk for ulcer is increased, but this Standards of Medical Commonly Available would not be our preferred wording, as Care in Diabetes– Clinical Information we feel certain about this association, hav- 2006 ing published data reporting this finding in 1996 (5). Comparison of cumulative Response to Leese and Morris risks, as in their report, may thus lead to Response to the American Diabetes biased results due to differential fol- Association low-up times by degree of risk. Fortu- e appreciate the interest of Leese nately, analytic methods that compare and Morris (1) in our article and recently encountered a discrepancy in enjoyed reading their publica- failure times and employ censoring, the American Diabetes Association’s W which we used in our recent publication, tion (2), which was published 1 month I (ADA’s) recommendations regarding before our own and hence was not cited can address this problem. conventional versus SI units for HDL cho- by us (3). We wish to point out that their Hopefully these recent efforts to de- lesterol (1). The article states “raise HDL characterization of our population as hav- velop foot ulcer prediction models and cholesterol to Ͼ40 mg/dl (1.15 mmol/l).” ing been “recruited from a hospital diabe- others will lead to more accurate identifica- Simple calculation shows that 40 mg/dl ϭ tes clinic” is not correct, as we clearly tion of individuals at higher risk of foot ulcer 40 ϫ 0.02586 mmol/l ϭ 1.03 mmol/l, describe that our subjects were recruited and stimulate the development of better rather than 1.15 mmol/l. The same error from a general internal medicine clinic preventive strategies to reduce morbidity is also noted in the 2005 version. (3). With regard to the generalizability of and mortality from these complications. Furthermore, in regard to the ADA’s our findings, they should be applicable to recommendation to use statin therapy for the ϳ25 million U.S. veterans and other 1,2 diabetic patients without overt cardiovas- EDWARD J. BOYKO, MD, MPH males with similar clinical and demo- 1 cular disease (CVD), the recommenda- JESSIE H. AHRONI, ARNP, PHD graphic characteristics enrolled in a pri- 1,2 tion to treat “regardless of baseline LDL” mary care setting. Although Leese and KARIN M. NELSON, MD might have extended beyond the evi- Morris write that they have “already ad- dence quoted. I find the recommenda- dressed” foot ulcer prediction in their From the 1Seattle Epidemiologic Research and Infor- tion’s evidence rather weak, despite a publication, they do not cite our publica- mation Center, VA Puget Sound Health Care System, similar recommendation elsewhere (2). tion from 1999, which presented a pre- Seattle, Washington; and the 2Department of Medi- For evidence on diabetic patients without diction model for diabetic foot ulcer and cine, University of Seattle, Seattle, Washington. overt cardiovascular problems, two stud- anticipated several of the findings of our Address correspondence to Edward J. Boyko, ies are listed (3,4). Here is an abbreviated MD, MPH, VA Epidemiologic Research and Infor- recent article (4). mation Center, 1100 Olive Way, Suite 1400, Seattle, synopsis: Leese and Morris describe their find- WA 98101. E-mail: [email protected]. The Heart Protection Study (3) ings as useful for “all-comers” in a general DOI: 10.2337/dc06-1661 showed cardiovascular benefit of statin community setting and in specialized © 2006 by the American Diabetes Association. therapy for LDL Ͼ3.0 mmol/l similar to

DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 2563 Letters

Ͻ3.0 mmol/l. Only 49% of the study’s © 2006 by the American Diabetes Association. mmol/l) for men and 50 mg/dl (1.29 diabetic subjects were free of overt CVD, mmol/l) for women. We will make this and 19% had prior myocardial infarction. ●●●●●●●●●●●●●●●●●●●●●●● change for the 2007 position statement. The CARDS (Collaborative Atorva- References Tseng also raises the issue that he statin Diabetes Study) (4) showed cardio- 1. American Diabetes Association: Stan- does not feel that the evidence supplied vascular benefit of statin therapy for LDL dards of medical care in diabetes–2006 supports our recommendation that “For Ͼ3.1 mmol/l or Ͻ3.1 mmol/l. Subjects (Position Statement). Diabetes Care 29 those over the age of 40 years, statin ther- were without CVD but had at least one of (Suppl. 1):S4–S42, 2006 apy to achieve an LDL reduction of 30– hypertension, retinopathy, proteinuria, 2. Williams B, Poulter NR, Brown MJ, Davis 40% regardless of baseline LDL is M, McInnes GT, Potter JF, Sever PS, or smoking. Thom SM: British Hypertension Society recommended.” We are in disagreement I believe the more appropriate inter- Guidelines for hypertension management on this point. While he is correct that in pretation of these two studies is that car- 2004 (BHS-IV): summary. BMJ 328:634– the Heart Protection Study (3), the cut diovascular benefits of the statin therapy 640, 2004 point for LDL level was 3.0 mmol/l (116 on diabetic patients have been shown for 3. Collins R, Armitage J, Parish S, Sleigh P, mg/dl), they found that “lowering the LDL LDL Ͻ or Ͼ3.1 mmol/l (or 3.0 mmol/l, Peto R, the Heart Protection Study Collab- cholesterol from Ͻ3.0 mmol/l to Ͻ2.0 depending on the study). However, the orative Group: MRC/BHF Heart Protec- mmol/l (i.e., Ͻ116–Ͻ77 mg/dl) in peo- precise lower limit of LDL, where the car- tion Study of cholesterol-lowering with ple with diabetes reduces macrovascular diovascular benefits may or may not per- simvastatin in 5,963 people with diabetes: disease” and concluded that “statin ther- sist, has not been explored. Using a a randomised placebo-controlled trial. apy should now be considered routinely Lancet 361:2005–2016, 2003 threshold of 3.0 mmol/l for LDL is not a 4. Colhoun HM, Betteridge DJ, Durrington for all diabetic patients at sufficiently high replacement for using successively lower PN, Hitman GA, Neil HA, Livingstone SJ, risk of such major vascular event, irre- thresholds. Using an allegory may help. Thomason MJ, Mackness MI, Charlton- spective of their initial cholesterol con- For example, one cannot claim that since Menys V, Fuller JH, the CARDS Investiga- centrations.” Similarly, while the cut antihypertensive therapy reduces cardio- tors: Primary prevention of cardiovascular point in the Coronary Artery Diabetes vascular events for patients with blood disease with atorvastatin in type 2 diabetes Study (4) was 3.1 mmol/l (120 mg/dl), the pressure of Ͼ160/100 mmHg, as well as in the Collaborative Atorvastatin Diabetes median LDL cholesterol on statin treat- Ͻ160/100 mmHg, antihypertensive Study (CARDS): multicentre randomized ment was 2.0 mmol/l (77 mg/dl), and 25% medications should be used regardless of placebo-controlled trial. Lancet 364:685– had a concentration Ͻ1.7 mmol/l (66 mg/ any baseline blood pressure. Our current 696, 2004 dl). The Coronary Artery Diabetes Study 5. Sever PS, Poulter NR, Dahlof B, Wedel H, ignorance of the lowest LDL threshold of Collins R, Beevers G, Caulfield M, Kjeld- investigators also concluded that “no jus- statin’s benefits should not be replaced by sen SE, Kristinsson A, McInnes GT, tification is available for having a particu- the statement that no lower limit to the Mehlsen J, Nieminen M, O’Brien E, Oster- lar threshold level of LDL cholesterol as benefit of LDL lowering exists. A recent gren J: Reduction in cardiovascular events the sole arbiter of which patients with study (5) not referenced in the position with atorvastatin in 2,532 patients with type 2 diabetes should receive statins.” statement showed cardiovascular benefit type 2 diabetes: Anglo-Scandinavian Car- Based on the above, we continue to feel of lowering LDL from 3.3 to 2.15 mmol/l, diac Outcomes Trial–Lipid-Lowering our recommendation is appropriate and but again, it does not provide an answer as Arm (ASCOT-LLA). Diabetes Care 28: evidence based. Of note, we do review to whether unlimited lowering of LDL 1151–1157, 2005 and revise the American Diabetes Associ- may cease to be useful (6). 6. Tseng KH: Reduction in cardiovascular ation Clinical Practice Recommendations events with atorvastatin in 2,532 patients Hence, I would recommend amend- with type 2 diabetes: Anglo-Scandinavian each year based on current evidence. ing the statement, “For those over the age Cardiac Outcomes Trial–Lipid-Lowering 1 of 40 years, statin therapy to achieve an Arm (ASCOT-LLA) (Letter). Diabetes Care VIVIAN FONSECA, MD 2 LDL reduction of 30–40%, regardless of 28:2595, 2005 NATHANIEL G. CLARK, MD, MS, RD baseline LDL levels, is recommended (A).” This can be replaced by “For those From the 1Division of Endocrinology, Department of Medicine, Tulane University Medical Center, over the age of 40 years, without overt Standards of Medical 2 CVD or cardiovascular risk factors other New Orleans, Louisiana; and the American Diabe- Care in Diabetes– tes Association, Alexandria, Virginia. than diabetes, statin therapy to lower LDL Address correspondence to Richard A. Kahn, from Ͼ3.0 mmol/l to Ͻ3.0 mmol/l is ad- 2006 PhD, American Diabetes Association, 1701 N. visable. The lowest baseline LDL where Beauregard St., Alexandria, VA 22311. E-mail: statin therapy might cease to be of benefit [email protected]. Response to Tseng DOI: 10.2337/dc06-1150 is currently unexplored and therefore un- © 2006 by the American Diabetes Association. known.” e appreciate Dr. Tseng’s letter ●●●●●●●●●●●●●●●●●●●●●●● KINGSTON H. TSENG, MBBS, FACP, FACE (1), as it identifies an error in our 2006 stated HDL cholesterol References W 1. Tseng KH: Standards of medical care in goals (2), when expressed in SI units, and From Pacific Medical Associates, Kowloon, Hong diabetes–2006 (Letter). Diabetes Care 29: Kong. allows us to comment further on our rec- 2563–2564, 2006 Address correspondence to Dr. Kingston H. ommendation regarding the use of statins Tseng, Endocrinologist, Pacific Medical Associates, 2. American Diabetes Association: Stan- Suite 1305, 102 Austin Rd., TST, Kowloon, Hong in those with type 2 diabetes. dards of medical care in diabetes–2006 Kong. E-mail: [email protected]. Tseng is correct that the 2006 goals (Position Statement) Diabetes Care 29 DOI: 10.2337/dc06-0805 for HDL (2) should read 40 mg/dl (1.03 (Suppl. 1):S4–S42, 2006

2564 DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 Letters

3. Heart Protection Study Collaborative events is no more than the sum of its miological studies, the short-term risk for Group: MRC/BHF Heart Protection Study parts. Whether the metabolic syndrome CVD for the syndrome as a whole, as- of cholesterol-lowering with simvastatin serves an important function to alert phy- sessed by current diagnostic criteria, does in 5,963 people with diabetes: a random- sicians and patients of the importance of not substantially exceed the risk con- ised placebo-controlled trial. Lancet 361: addressing these risk factors is a separate tained in two of its risk factors. Conse- 2005–2016, 2003 4. Colhoun HM, Betteridge DJ, Durrington issue. quently, from a predictive perspective, PN, Hitman GA, Neil HA, Livingstone SJ, the metabolism syndrome is really noth- Thomason MJ, Mackness MI, Charlton- MAYER B. DAVIDSON, MD ing more than a higher blood pressure Menys V, Fuller JH: Primary prevention of and a lower HDL. The argument from cardiovascular disease with atorvastatin From the Clinical Center for Research Excellence, causation holds that these predictors of in type 2 diabetes in the Collaborative Charles R. Drew University, Los Angeles, California. risk do not necessary reflect the true Atorvastatin Diabetes Study (CARDS): Address correspondence to Mayer B. Davidson, MD, Clinical Center for Research Excellence, causes of the risk. Indeed, there are other multicentre randomised placebo-con- Charles R. Drew University, 1731 E. 120th St., Los types of data to indicate that several risk trolled trial. Lancet 364:685–696, 2004 Angeles, CA 90059. E-mail: mayerdavidson@ factors of the syndrome, such as eleva- cdrewu.edu. tions in VLDL, a prothromobotic state DOI: 10.2337/dc06-1592 Does the Metabolic © 2006 by the American Diabetes Association. and a proinflammatory state also contrib- ute to risk. It is possible to say that for Syndrome Exist ●●●●●●●●●●●●●●●●●●●●●●● causation, robust predictors are con- founded by other risk factors, some of References Response to Grundy 1. Grundy SM: Does the metabolic syn- which are not routinely measured. In fact, drome exist? (Commentary). Diabetes there is still uncertainty as to whether a Care 29:1689–1692, 2006 low HDL cholesterol is truly a direct cause key question in the debate concern- 2. Kahn R: The metabolic syndrome (em- of CVD or is only a marker for risk. Fur- ing the metabolic syndrome (1,2) is peror) wears no clothes (Commentary). ther, clinical trials indicate that reducing A whether the risk accompanying it is Diabetes Care 29:1693–1696, 2006 blood pressure does not fully reverse the more than the sum of its parts. Grundy (1) 3. Yusuf S, Hawken S, Ounpuu S, Bautista L, risk predicted by a higher blood pressure; asserts that “risk factors are multiplica- Franzosi MG, Commerford P, Lang CC, hence risk associated with a higher blood tive, i.e., risk for ASCVD [atherosclerotic Rumboldt Z, Onen CL, Lisheng L, Tan- pressure must be confounded by other cardiovascular disease] from risk factors omsup S, Wangai P Jr, Razak F, Sharma factors. The clinical implications of this AM, Anand SS, the INTERHEART Study rises geometrically, not linearly, as the Investigators: Obesity and the risk of distinction between prediction and cau- number of risk factors increases. There- myocardial infarction in 27,000 partici- sation are considerable. At present, it can- fore, total risk is more than a summation pants from 52 countries: a case-control not be assumed that treatments directed of the individual factors.” The study by study. Lancet 366:1640–1649, 2005 toward the predictors will produce the Yusuf et al. (3) is offered to support this 4. Sunderstro¨m J, Vallhagen E, Rise´rus U, expected reduction in risk; rather, it is statement, but this study is a standardized Byberg L, Zethelius B, Berne C, Lind L, important to identify the true causes of case-control study of 27,098 participants Ingelsson E: Risk associated with the met- CVD associated with the metabolic syn- in 52 countries representing several major abolic syndrome versus the sum of its in- drome so that they can be better targeted ethnic groups to assess the relation be- dividual components. Diabetes Care 29: for therapy. tween BMI, waist and hip circumferences, 1673–1674, 2006 A second line of debate is whether the and waist-to-hip ratio to myocardial in- risk factors of the metabolic syndrome, or farction overall and for each group. The Does the Metabolic indeed for all CVD, are additive or syner- metabolic syndrome is not even men- gistic in their effect on CVD risk. Synergism tioned in the article. Syndrome Exist? in effect is referred to as multiplicative In the same issue of Diabetes Care, risk. Several epidemiological studies and Sunderstro¨m et al. (4) published a study Response to Davidson the risk algorithms developed from these evaluating the risk factors for cardiovas- studies support the synergism associated cular death in Ͼ2,000 individuals fol- with multiple risk factors. If multiple risk lowed for 30 years after being studied at n response to Davidson (1), who ques- factors are synergistic in their effects, as age 50. More than 1,000 of them were tions whether the metabolic syndrome epidemiology indicates, then the risk as- reexamined at age 70 and followed for 9 I conveys more risk for cardiovascular sociated with multiple risk factors is more years. Sunderstro¨m et al. showed disease (CVD) than that contained in its greater than what would be obtained by that the “metabolic syndrome did not pre- risk components, I suggest that this de- simple addition of their individual effects. dict cardiovascular mortality indepen- bate is partly confused by ambiguity over Finally, the metabolic syndrome is a dently of its individual components at any the issues of prediction versus causation. progressive condition. It typically wors- age” and concluded that “the metabolic The argument based on prediction con- ens with advancing age. Hence, risk is syndrome might be viewed as a clinically tends that among the diagnostic compo- compounded by aging. This means that handy summary measure of nontradi- nents of the syndrome (elevated blood risk predicted at any one time underesti- tional risk factors rather than as a strong pressure, low levels of HDL cholesterol, mates the long-term risk associated with biological entity.” Thus, this evidence high triglycerides, elevated glucose, and the syndrome. suggests that the answer to the key ques- abdominal obesity), most of the predic- These multiple lines of evidence sup- tion posed above is that the risk of the tive power is contained in blood pressure port my contention that the CVD risk ac- metabolic syndrome for cardiovascular and HDL cholesterol. Therefore, in epide- companying the metabolic syndrome

DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 2565 Letters

cannot be causally explained by the fac- mal BMI (excluding those with diabetes). bolic syndrome, diabetes, and prevalence tors contained in its current diagnostic Reaven (2) pointed out that only about of coronary heart disease among criteria. The latter serves to identify the one-third of the most insulin-resistant in- NHANES III participants age 50 years and presence of higher risk condition but does dividuals were actually obese, and the de- older. Diabetes 52:1210–1214, 2003 not necessarily represent the sole targets gree of correlation between insulin- 7. McNeill AM, Rosamond WD, Girman CJ, Golden SH, Schmidt MI, East HE, Ballan- of therapy for the condition. mediated glucose uptake and BMI, waist tyne CM, Heiss G: The metabolic syn- circumference, and visceral obesity were drome and 11-year risk of incident 1,2,3 SCOTT M. GRUNDY, MD, PHD almost equal. Neither abdominal obesity cardiovascular disease in the Atheroscle- nor metabolic syndrome, as defined by rosis Risk in Communities study. Diabetes From the 1Center for Human Nutrition, University the National Cholesterol Education Pro- Care 28:385–390, 2005 of Texas Southwestern Medical Center, Dallas, Tex- 2 gram Adult Treatment Panel III criteria, 8. Ridker PM, Cook N: Clinical usefulness of as; the Department of Clinical Nutrition and Inter- very high and very low levels of C-reactive nal Medicine, University of Texas Southwestern were a significant independent risk factor Medical Center, Dallas, Texas; and the 3Department for CVD in multiple regression analyses protein across the full range of Framing- of Internal Medicine, University of Texas Southwest- (6,7), while very low levels of CRP were ham Risk Scores. Circulation 109:1955– ern Medical Center, Dallas, Texas. useful for risk prediction among women 1959, 2004 9. Oda E, Oohara K, Abe A, Veeraveedu PT, Address correspondence to Scott M. Grundy, with calculated 10-year Framingham MD, PhD, Center for Human Nutrition and Depart- Ͻ Watanabe K, Kato K, Aizawa Y: The opti- ments of Clinical Nutrition and Internal Medicine, risks 10% (8). Previously, we propose a mal cut-off point of C-reactive protein as University of Texas Southwestern Medical Center, CRP value of 0.65 mg/l as the cut point for an optional component of metabolic syn- 5323 Harry Hines Blvd., Y3.206, Dallas, TX 75390- metabolic syndrome (9) instead of ethnic- drome in Japan. Circ J 70:384–388, 2006 9052. E-mail: [email protected]. specific controversial cut points of waist DOI: 10.2337/dc06-1725 © 2006 by the American Diabetes Association. circumference. Whether the new clothes fit the emperor should be revealed by re- The Metabolic ●●●●●●●●●●●●●●●●●●●●●●● analyses of existing epidemiological stud- Syndrome (Emperor) ies including CRP data. References Wears No Clothes 1. Davidson MB: Does the metabolic syn- drome exist? (Letter). Diabetes Care 29: EIJI ODA, MD 2565, 2006 Response to Oda From the Department of Internal Medicine, Niigata Prefectural Yoshida Hospital, Niigata, Japan. Address correspondence to Eiji Oda, MD, De- The Metabolic partment of Internal Medicine, Niigata Prefectural he possible inclusion of C-reactive Syndrome (Emperor) Yoshida Hospital, Yoshida-Daibochou 32-14, Tsub- protein (CRP) in the metabolic syn- ame, Niigata, 959-0242, Japan. E-mail: ijie@ T drome algorithm highlights the Wears No Clothes venus.sannet.ne.jp. problems of the construct itself. If the syn- DOI: 10.2337/dc06-1453 drome is supposed to be a predictive tool © 2006 by the American Diabetes Association. Response to Kahn for future cardiovascular events, then we ●●●●●●●●●●●●●●●●●●●●●●● should indeed test the benefit of adding CRP, along with age, sex, race, adiponec- References ahn (1) pointed out that metabolic tin, homocysteine, etc. If the utility of the 1. Kahn R: The metabolic syndrome (em- syndrome is to call attention to obesity, syndrome is associated with many peror) wears no clothes (Commentary). K uncertainties and inconsistencies, Diabetes Care 29:1693–1696, 2006 then there is no need for expensive lab- which could easily misdirect care, mis- 2. Reaven GM: The metabolic syndrome: is oratory tests. If the virtue is to predict lead patients, and lead to unnecessary this diagnosis necessary? Am J Clin Nutr diabetes, then a measure of glucose intol- health care costs. The risk of cardiovascu- 83:1237–1247, 2006 erance alone is better. If the syndrome is lar disease (CVD) associated with the syn- 3. Shamsuzzaman AS, Winnicki M, Wolk R, supposed to identify those with insulin drome is no greater than that explained by Svatikova A, Phillips BG, Davison DE, resistance, then there are simpler ways to the presence of its components, and it is Berger PB, Somers VK: Independent asso- do so; however, measuring CRP might be possible to create an almost infinite num- ciation between plasma leptin and C-re- helpful. active protein in healthy humans. ber of scenarios in which individuals who Dr. Oda (1) seems to have one pur- Circulation 109:2181–2185, 2004 pose for the construct; others have differ- do not meet the diagnostic criteria for 4. Bo S, Gambino R, Uberti B, Mangiameli metabolic syndrome would be at greater MP, Colosso G, Repetti E, Gentile L, Cas- ent perspectives. The problem is that no risk of CVD than would those who do, as sader M, Pagano GF: Does C-reactive pro- one, not even the proponents themselves noted by Reaven (2). tein identify a subclinical metabolic (2–4), have conveyed the exact utility of What if, among the diagnostic com- disease in healthy subjects? Eur J Clin In- the syndrome and shown that it is better ponents of metabolic syndrome, waist vest 35:265–270, 2005 than or even equal to other approaches, circumference, which is one of the an- 5. Matsushita K, Yatsuya H, Tamakoshi K, and the critical issue, therefore, is why thropometric markers of obesity, was Wada K, Otsuka R, Zhang H, Sugiura K, clinicians should even bother diagnosing substituted by high-sensitivity C-reactive Kondo T, Murohara T, Toyoshima H: In- metabolic syndrome in the first place. verse association between adiponectin protein (CRP)? CRP is a sensitive marker and C-reactive protein in substantially RICHARD KAHN, PHD of subclinical systemic inflammation and healthy Japanese men. Atherosclerosis 188: positively relates to leptin (3) and insulin 184–189, 2006 From the American Diabetes Association, Alexan- resistance (4) and negatively relates to 6. Alexander CM, Landsman PB, Teutsch dria, Virginia. adiponectin (5), even in people with nor- SM, Haffner SM: NCEP-defined meta- Address correspondence to Richard Kahn, PhD,

2566 DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 Letters

American Diabetes Association, 1701 N. Beauregard cope with the possible fact that on aver- diabetes increases accumulation of POPs St., Alexandria, VA 22311. E-mail: rkahn@diabetes. age, diabetic subjects have higher concen- seems scarce (4). Do the authors know of org. DOI: 10.2337/dc06-1616 trations of POPs and may thus be more studies on the toxicokinetics of POPs in © 2006 by the American Diabetes Association. likely to suffer the adverse effects of POPs. diabetic subjects demonstrating that they The mechanistic, clinical, and public accumulate POPs more than nondiabetic ●●●●●●●●●●●●●●●●●●●●●●● health implications of the study by Lee et al. subjects? are potentially high (1,3–5). However, sev- References eral questions remain unanswered regard- 1. Oda E: The metabolic syndrome (em- MIQUEL PORTA, MD, MPH, PHD peror) wears no clothes (Letter). Diabetes ing the nature of the relationship between Care 29:2566, 2006 prevalence of diabetes and population From the School of Medicine, Universitat Auto`noma 2. Alberti KG, Zimmet P, Shaw J: The meta- distribution of POPs (6,7). Therefore, I de Barcelona and Institut Municipal d’Investigacio´ bolic syndrome: a new worldwide defini- would appreciate it if Lee et al. could ad- Me`dica, Barcelona, Catalonia, Spain. tion. Lancet 366:1059–1062, 2005 dress the following issues. Address correspondence to Prof. Miquel Porta, School of Medicine, Universitat Auto`noma de Bar- 3. Grundy SM, Brewer HB Jr, Cleeman JI, 1) What is the direction of the rela- celona and Institut Municipal d’Investigacio´Me`dica, Smith SC Jr, Lenfant C: Definition of met- tionships with the poverty income ratio? Carrer del Dr Aiguader 80, E-08003 Barcelona, abolic syndrome: report of the National For example, in Table 1, did wealthier in- Catalonia, Spain. E-mail: [email protected]. Heart, Lung, and Blood Institute/Amer- dividuals have lower concentrations of DOI: 10.2337/dc06-1531 ican Heart Association conference on scien- DDE and higher concentrations of © 2006 by the American Diabetes Association. tific issues related to definition. Circulation 109:433–438, 2004 PCB153 after adjusting for confounders? ●●●●●●●●●●●●●●●●●●●●●●● 4. Grundy SM: Does the metabolic syndrome 2) Many of the estimates (e.g., in Ta- exist? (Commentary). Diabetes Care 29: ble 2) were adjusted for age, sex, race, References 1689–1692, 2006 income, lipids, BMI, and waist circumfer- 1. Lee D-H, Lee I-K, Song K, Steffes M, To- ence. This is coherent with several aims scano W, Baker BA, Jacobs DR Jr: A strong (e.g., to “isolate” the effect of POPs from dose-response relation between serum A Strong Dose- that due to obesity, age, race, or income). concentrations of persistent organic pol- However, adjusting by BMI and waist cir- lutants and diabetes: results from the Na- Response Relation cumference may also be an overadjust- tional Health and Examination Survey 1999–2002. Diabetes Care 29:1638– Between Serum ment, since fat intake is the most common 1644, 2006 Concentrations of source of exposure to POPs (1,5) and 2. Pladevall M, Singal B, Williams LK, Bro- since the body burden of some of these tons C, Guyer H, Sadurni J, Falces C, Ser- Persistent Organic lipophilic chemicals, but not all and not rano-Rios M, Gabriel R, Shaw JE, Zimmet Pollutants and always, increases with increasing BMI PZ, Haffner S: A single factor underlies the Diabetes: Results (8,9). Thus, crude or less adjusted odds metabolic syndrome: a confirmatory fac- ratios (ORs) would also be relevant for tor analysis. Diabetes Care 29:113–122, From the National determining the prevalence of diabetes in 2006 Health and Nutrition people with specific concentrations of 3. Vasiliu O, Cameron L, Gardiner J, DeGuire POPs. Could the authors provide some P, Karmaus W: Polybrominated biphenyls, Examination Survey polychlorinated biphenyls, body weight, crude ORs? and incidence of adult-onset diabetes 1999–2002 3) The finding that there was no asso- mellitus. Epidemiology 17:352–359, 2006 ciation between obesity and diabetes 4. Longnecker MP: Pharmacokinetic vari- Response to Lee et al. among subjects with nondetectable levels ability and the miracle of modern analyt- of POPs is highly surprising and calls for ical chemistry. Epidemiology 17:350–351, additional results to be presented. A fig- 2006 ee et al. (1) and Diabetes Care deserve ure may be warranted. 5. Porta M: Persistent organic pollutants and praise for publishing what may be 4) Also crucial is what we may call the burden of diabetes. Lancet 368:558– the first study worldwide to analyze, “the changes in BMI-POPs relationship.” 559, 2006 L 6. Rewers M, Zimmet P: The rising tide of in a sample of a general population, serum Could the authors please comment on the childhood type 1 diabetes: what is the elu- concentrations of persistent organic pol- possible influence upon BMI measure- sive environmental trigger? Lancet 364: lutants (POPs) and plasma fasting glu- ments of the cross-sectional design of the 1645–1647, 2004 cose. The main implication of the study is study? Could they suggest possible con- 7. Daneman D: Type 1 diabetes. Lancet 367: that POPs stored in the adipose tissue may sequences upon findings of weight gain 847–858, 2006 be a key player in the etiopathogenesis of and weight loss in diabetic and nondia- 8. Wolff MS, Britton JA, Teitelbaum SL, Eng type 2 diabetes. It is even rational to spec- betic participants? S, Deych E, Ireland K, Liu Z, Neugut AI, ulate that POPs might be, if not “the single 5) High-prevalence ORs were found Santella RM, Gammon MD: Improving factor” (2), then one factor linking some for the summary or composite of the six organochlorine biomarker models for core components of the metabolic POPs with the highest concentrations. cancer research. Cancer Epidemiol Biomar- syndrome. Are the results similar if the joint effects of kers Prev 14:2224–2236, 2005 9. Perry MJ, Ouyang F, Korrick S, Venners In the study by Lee et al. and other multiple POPs are assessed through alter- SA, Altshul L, Xu X, Wang X: Body mass studies (3,4), it seems likely that a rela- native methods? index and serum 1,1,1-trichloro-2,2- tionship exists between diabetes and 6) Finally, the authors state that “re- bis(p-chlorophenyl)ethane in nulliparous POPs. Hence, patients, clinicians, and verse causality is unlikely.” Indeed, evi- Chinese women. Cancer Epidemiol Bio- other health professionals may need to dence supporting the hypothesis that markers Prev 14:2433–2438, 2005

DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006 2567 Letters

A Strong Dose- tion to the possibility of reverse causality From the 1Department of Preventive Medicine and because BMI could be inversely related to Health Promotion Research Center, School of Med- Response Relation icine, Kyungpook National University, Daegu, Ko- clearance of POPs (4) and since diabetic rea; the 2Division of Epidemiology, School of Public Between Serum patients are more obese. Thus, diabetic Health, University of Minnesota, Minneapolis, Min- Concentrations of patients may increase accumulation of nesota; and the 3Department of Nutrition, Univer- POPs in their body. However, one study sity of Oslo, Oslo, Norway. Persistent Organic reported no significant difference be- Address correspondence to Duk-Hee Lee, MD, PhD, Department of Preventive Medicine, School of Pollutants and tween the average elimination rates of di- Medicine, Kyungpook University, 101 Dongin- Diabetes: Results abetic and nondiabetic individuals (5). In dong, Jung-gu, Daegu, Korea 700-422. E-mail: addition, we adjusted for both BMI and [email protected]. From the National waist circumference. Thus, decreased DOI: 10.2337/dc06-1720 Health and Nutrition elimination does not appear to explain © 2006 by the American Diabetes Association. Examination Survey our finding. We agree that one of the most surpris- ●●●●●●●●●●●●●●●●●●●●●●● 1999–2002 ing results would be that the prevalence of diabetes itself was quite low and that obe- References 1. Porta M: A strong dose-response relation sity was not associated with diabetes Response to Porta between serum concentrations of persis- among subjects with very low levels of tent organic pollutants and diabetes: re- POPs, suggesting that POPs contained in sults from the National Health and e thank Dr. Porta (1) for his in- the adipose tissue, not obesity, may be a Examination Survey 1999–2002 (Letter). terest in our study (2). As our key to diabetes. Some of our findings may Diabetes Care 29:2567, 2006 study was cross-sectional, a pos- be criticized due to the cross-sectional de- 2. Lee D-H, Lee I-K, Song K, Steffes M, To- W scano W, Baker BA, Jacobs DR Jr: A strong sibility of reverse causality needs to be sign. However, the lack of association be- tween obesity and diabetes among dose-response relation between serum carefully evaluated. The most critical is- concentrations of persistent organic pol- sue he raised is weight changes, which subjects with very low levels of POPs is unlikely to be a cross-sectional bias. Al- lutants and diabetes: results from the Na- were not considered in our report. Body though Diabetes Care does not allow fig- tional Health and Examination Survey weight loss increases both serum and ad- 1999–2002. Diabetes Care 29:1638– ures in response letters, the requested ipose tissue concentrations of persistent 1644, 2006 figure depicting the interaction between organic pollutants (POPs) (3,4), and dia- 3. Chevrier J, Dewailly E, Ayotte P, Mauriege obesity and POPs can be drawn from our P, Despres JP, Tremblay A: Body weight betic patients are advised to attempt publication and looks very persuasive. weight loss as a nonpharmacologic inter- loss increases plasma and adipose tissue Regarding Dr. Porta’s other ques- concentrations of potentially toxic pollut- vention. Conceivably, the strong relation tions, wealthier individuals (higher ants in obese individuals. Int J Obes Relat between POPs and prevalence of diabetes poverty income ratio) had lower concen- Metab Disord 24:1272–1278, 2000 could be explained by weight loss among trations of DDE but higher concentrations 4. Wolff MS, Britton JA, Teitelbaum SL, Eng diabetic patients. The National Health of PCB153. ORs adjusted only for age S, Deych E, Ireland K, Liu Z, Neugut AI, and Nutrition Examination Survey col- were 14.4, 18.3, 45.5, and 47.0, with lit- Santella RM, Gammon MD: Improving lected information on body weight 1 and tle change on further adjustment for sex organochlorine biomarker models for cancer research. Cancer Epidemiol Biomar- 10 years before examination. Additional and race/ethnicity. Several methods of adjustment for weight change over the kers Prev 14:2224–2236, 2005 combining the six POPs yielded conclu- 5. Michalek JE, Ketchum NS, Tripathi RC: past 1 or 10 years attenuated the original sions similar to those we presented. Diabetes mellitus and 2,3,7,8-tetrachlo- odds ratios (ORs) to 12.8, 10.8, 26.5, and rodibenzo-p-dioxin elimination in veter- 1 26.7 or 9.1, 6.7, 16.0, and 15.8. BMI itself DUK-HEE LEE, MD, PHD ans of Operation Ranch Hand. J Toxicol 2,3 may be another important issue in rela- DAVID R. JACOBS,JR., PHD Environ Health A 66:211–221, 2003

2568 DIABETES CARE, VOLUME 29, NUMBER 11, NOVEMBER 2006