meta分析范文展示

1、9The effect of fructose consumption on plasma cholesterol in adults: a meta-a nalysis of con trolled feedi ng trials 1,2,3Tao An4,5, Rong Cheng Zhan&,5, Yu Hui Zhang4, Qio ng Zhou4, Yan Huang Jia n Zhang4, *.4 State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardio

2、vascular Diseases, Chin ese Academy of Medical Scie nces and Pek ing Un io n Medical College, Beiji ng, Chi na5 Tao An and Rong Cheng Zhang con tributed equally to this study.3Supplemental Table 1 and supplemental Figures 1-4 are available as Online Supporting Material with the online posting of thi

3、s paper atRUNNING TITLE: Fructose and cholesterolWORD COUNT: 5618; NUMBER OF FIGUREA: 3; NUMBER OF TABLES: 2SUPPLEMENTARY MATERIAL: On li ne Supporti ng Materials: 5AUTHOR LIST FOR INDEXING: An, Zha ng, Zha ng, Zhou, Hua ng, Zha ng1The study was supported by the Mini stry of Scie nee and Tech no log

4、y of China with grant of the National High-tech Research and Development Program of China to Dr Jia n Zhang.2Author disclosures: T. An, R.C. Zhang, YH. Zhang, Q. Zhou, Y Hung, J. Zhang have no con flicts of in terest.* To whom corresp ondence should be addressed. Maili ng address: Heart Failure Cen

5、ter, Cardiovascular In stitute and Fuwai Hospital, Chin ese Academy of Medical Scien ces and Peking Un io n Medical College, 167 Beilishilu, Beiji ng, Chi na; Zip code: 100000; Telephone number: 86-10-88396180; Fax number: 86-10-; E-mail:PROSPERO REGISTRATION NUMBERS: CRD4201200335119 / 319234567891

6、011121314151617181920212223ABSTRACTFructose is widely used as a sweete ner in product on of many foods, yet the relati on betwee n fructose in take and cholesterol remai ns un certa in. We performed a systematic review and meta-a nalysis of huma n con trolled feed ing trials of isocaloric fructose e

7、xchange for other carbohydrates to quantify the effects of fructose on total cholesterol (TC), LDL cholesterol (LDL-C), and HDL cholesterol (HDL-C) in adult humans. Weighted mean differences were calculated for changes from baseline cholesterol concen trati ons by using gen eric inv erse varia nee r

8、an dom-effects models. The Heyland Methodological Quality was used to assessstudy quality. Subgroup an alyses and meta-regressi on were con ducted to explore possible in flue nee of study characteristics. Twen ty-four trials (with a total of 474 subjects) were in cluded in our meta-a nalysis. In an

9、overall pooled estimate, fructose exerted no effect on TC, LDL-C and HDL-C. Meta-regression analysis indicated that fructose dose was positively correlated with the effect sizes of TC and LDL-C. Subgroup analyses showed that isocaloric fructose excha nge for carbohydrates could sig nifica ntly in cr

10、ease TC by 12.97 mg/dL (95%CI: 4.66, 21.29; P = 0.002) and LDL-C by 11.59 mg/dL (95%CI: 4.39, 18.78; P = 0.002) at 100g fructose/d but had no effect o n TC and LDL-C whe n fructose in take was 100g/d) could lead to sig nifica ntly in crease in serum LDL-C and TC. Larger, l on ger and higher-quality

11、huma n con trolled feed ing trials are n eeded to confirm these results.Key words: fructose, cholesterol, meta-a nalysis24252627282930313233343536373839404142434445INTRODUCTIONHyperlipidemia is a com mon risk factor for coronary heart disease (CHD), with 44.4% of adults in the United States having a

12、bnormal TC values and 32% having elevated LDL-C levels (1). Compared to subjects with no rmal blood lipid, those with hyperlipidemia have a 3-fold risk of heart attacks (2). Lifestyle modificati on should be in itiated in conjunction both primary and sec on dary preve nti on of CHD. More con siderat

13、i on exists as to what con stitutes healthy eat ing.Fructose is the most n aturally occurri ng mono saccharide, and has become a major con stitue nt of our moder n diet. Fruit, vegetables, and other n atural sources provide nearly one-third of dietary fructose, and two-thirds come from beverages and

14、 foods in the diets (eg, can dies, jam, syrups, etc) (3). Fructose is preferred by many people, especially those with diabetes mellitus because of its low glycemic in dex (23% versus glucose 100%) (4). After in testi nal uptake, fructose is mai nly removed from the blood stream by the liver in an in

15、sulin-independent manner, and is used for intrahepatic product ion of glucose, fatty acids or lactate. Cross-secti onal studies in huma n suggest that excessive fructose con sumpti on can lead to adverse metabolic effects, such as dyslipidemia and in creased visceral adiposity (5-7). The Dietary Gui

16、deli nes for Americans, 2010, point out that it is lack of sufficie nt evide nee to set a tolerable upper in take of carbohydratesfor adults (8). Although The Can dia n Diabetes Associati on suggests con sumptio n of no more tha n 60g of added fructose per day by people with diabetes for its triglyc

17、eride-raisi ng effect (9), the threshold dose of fructose at which the adverse in flue nee on cholesterol is con troversial.46474849505152535455565758596061626364656667To determine the effect of fructose on cholesterol, a substantial number of clinical trials have been performed on adult humans with

18、 different health status (diabetic, obese, overweight, hyperi nsuli nemic, impaired glucose-tolera nt and healthy). These trials used various in take levels of fructose and differe nt protocols. Thus, it is difficult to reach a con siste nt con clusi on across these studies. Therefore, we con ducted

19、 a systematic review of the scie ntific literature and meta-a nalysis of con trolled feedi ng trials to evaluate the effect of isocaloric oral fructose excha nge for carbohydrates on cholesterol and to clarify the active factors of fructose.Materials and MethodsThis meta-analysis followed the Prefer

20、red Reporting Items for Systematic Reviews and Meta-a nalyses (PRISMA) criteria (10).Search strategy. We searched PubMed (http:/; from 1966 to December 2012), Embase (http:/; from 1966 to December 2012) and the Cochra ne Library database () by using the following search terms: fructose and (lipemia

21、or lipaemia or lipids or cholesterol or total cholesterol” or LDL cholesterol or HDL cholesterol ) in En glish. We also searched China Natio nal Kno wledge In frastructure () and Wangfang database () in Chin ese accord ing to the search strategy. The search was restricted to reports of trials on hum

22、a ns.Study select ion. All cli ni cal trials using fructose and in dexed with in the above databaseswere collected. Two independent reviewers (T.A., R.C.Z) screened the abstracts and titles for initial inclusion. If this was not sufficient, full texts articles were obta ined and reviewed by at least

23、 two in depe ndent reviewers (T.A., R.C. Z, Q.Z.,Y.H.). The refere nee lists of retrieved articles also used to suppleme nt the database. Any disagreeme nts were resolved through discussi on. We in cluded con trolled feedi ng trials investigating the chronic effect of fructose on blood cholesterol,

24、from both randomized and nonrandomized studies, if they met the following criteria: subjects must have been administered fructose for at least 2 weeks; studies investigated the effect of oral free (unbound, mono saccharide) fructose whe n compared with isocaloric con trol diet with ano ther carbohyd

25、rate in place of fructose; studies were performed in human adults with either a parallel or crossover design; subjects in both experimental groups and control groups were instructed to consume isocaloric diets. If the study reported any comparis ons, we in cluded all such comparis ons in the meta-a

26、nalysis. Data extract ion and quality assessme nt. Two reviewers (T.A., R.C.Z) in depe nden tly extracted releva nt data from eligible studies. Disagreeme nts were resolved by one of the two authors (Y.H.Z., J.Z.). These data included information on study features (author, year of publication, study

27、 design, randomization, blinding, sample size, comparator, fructose form, dose, follow-up and macronu trie nt pro the backgro und diet), participant characteristics (gender, age and healthy status) and baseline and final concentrations or net changes of total cholesterol, LDL-C and HDL-C. Data initi

28、ally extracted were con verted to system intern ati onal unit (eg, TC: 1 mmol/L conv erted to 38.6 mg/dL). For multi-arm studies, only intervention groups that met inclusion criteria were used in this analysis. If blood lipid concentrations were measured several times at different stages of trials,

29、only final records of lipid concentrations at the end of the trials were extracted for this meta-a nalysis.6869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111The quality of each study was assessed with the Heyla nd Methodological Quality Score (MQS) (

30、11), generalized as follows: randomization; analysis; blinding; patient selecti on; comparability of groups at baseli ne; exte nt of follow up; treatme nt protocol; co-i nterve nti on; outcomes. The highest score for each area was two poin ts. Higher nu mbers represe nted a better qualityMQS 8).Data

31、 synthesis. Statistical analyses were performed with Stata software (version 11.0; StataCorporation, TX, USA) and REVMAN software (version 5.2; Cochrane Collaborati on, Oxford, Un ited Kin gdom). Separatepooled an alyses were con ducted by using the gen eric in verse varia nee ran dom-effects models

32、 eve n where there was no evidenee of between-study heterogeneity because these models give more conservative summary effect estimates in the presenee of undetected residual heteroge neity tha n fixed-effects models. The differe nt cha nges from baseli ne betwee n fructose and carbohydrate comparato

33、rs for total cholesterol, LDL cholesterol and HDL cholesterol were used to estimate the principle effect. We applied paired an alyses to all crossover trials accord ing to the methods of Elbo urne and colleagues (12). Weighted mean differences of fructose consumption on cholesterol concen tratio ns

34、and corresp onding 95% CIs were calculated. A 2-sidedP value 0.05 was set as the level of significanee for an effect. The variances for net changes in serum cholesterol were only reported directly in two trials (29, 31). We calculate net changes for other studies by using the means土 SDs cholesterol

35、concentrations at baseline and at the end of intervention period (13). SDs were calculated from SEs when they were not directly given. If these data were unavailable, we extrapolatedmissing SDs by borrowing SDs derived from other trials in this meta-analysis (14). In addition, we assumed a conservat

36、ive degree of correlation of 0.5 to impute the change-from-baseline SDs, with sensitivity analyses performed across a range of possible correlatio n coefficie nts (0.25 and 0.75) (13). For crossover trials in which on ly final measureme nts were in cluded, the differe nces in mea n final measureme n

37、ts were assumedo n average to be the same as the differe nces in mean cha nge scores (13). Inter-study heterogeneity was tested by the Cochran Q-test (P 50% was evidenee of substantial heterogeneity. To explore the pote ntial effects of factors on the primary outcomes and inv estigate the possible s

38、ources of heterogeneity, we performed meta-regressionsand predefined subgroup analyses stratified by comparator, dose, study duration, randomization, health status, study desig n and study quality. As for studies used a range of fructose doses, the average doses calculated on the basis of the averag

39、e reported en ergy in take or weight of participants (28.5 calories per kilogram of body weight). Sensitivity analyses were also performed according to the Cochrane Handbook for Systemic Review Funnel plots and Egger snear regressi on test were con ducted to detect publicati on bias.RESULTSBased on

40、our search criteria, 1602 eligible studies were identified, and 1565 studies were excluded on review of the titles and abstracts. The remai ning 37 studies were retrieved and fully reviewed. Fiftee n of these did not meet the in clusi on criteria and were excluded in the final analysis. A total of 2

41、2 studies (providing data for 24134135136137138139140141142143144145146147148149150151152153154155trials) inv olvi ng474 subjects (15-36) were in eluded in the meta-a nalysis(Suppleme ntal Fig. 1, Table 1).The reports of Koh and Reiser (22, 23) in eluded two trials (bri nging the total number of tri

42、als to 24). Eleven trials were ran domized (17, 18, 20, 21,25, 27-29, 31, 34, 36). Ni netee n trials used crossover (15-19, 21-32), and five used parallel desig ns (20, 33-36). As for the 19 cross-over trials, 10 trials have reported the washout period (16, 18, 22, 25, 27-31), 9 trials did not have

43、washout period (15, 17, 19, 21,23, 24, 26, 32). The trials varied in size, from 8 to131 subjects. The mea n age of trial participa nts ran ged from 26.7 to 64.4 years. Seve nteen trials (15, 17-23, 25, 27, 28, 30, 31,34, 36) were performed in outpatie nt sett in gs, 3 trials (26, 29, 32) in in patie

44、 nt sett in gs, and 4 trials in both outpatient and inpatient settings (16, 24, 33, 35). Nine trials were con ducted on diabetic subjects (19-21,24-27, 29, 30), 8 trials in healthy subjects (17, 18, 22, 23, 28, 31,34, 35), 3 trials in overweight/obese subjects (32, 33, 36), 2 trials in hyperinsuline

45、mic subjects (16, 23),1 trial in those who were impairedglucose-tolera nt (22), and 1 trial in subjects with type IV hyperlipoprote in aemia (HLP) (15). Backgrou nd diets were 42-55% carbohydrate, 25-38% fat, and 13-20% protei n.The carbohydrate comparators choose starch in 13 trials (15, 16, 21,23-

46、25, 27-30, 32, 36), glucose in 6 trials (22, 31, 33-35), sucrose in 3 trials (17, 18, 26), and mixed carbohydrates in two trials (19, 20). Four trials used fructose in crystalline (16, 18, 20, 21), 5 trials in liquid (19, 32-35), and 15 trials in mixed form (15, 17, 22-31). The reported mean baselin

47、e serum TC ranged from 170 to 230.8 mg/dl, LDL-C ranged from 90.7 to 157 mg/dl, and HDL-C ran ged from 35.1 to 57.1 mg/dl. Nin eteen trials 11reported the fructose in take among backgro und diet was not differe nt betwee n the fructose and con trol groups, in which 15 trials reported the backgro und

48、 fructose in take account for 3% of total energy (9 to 24g) (15-23, 29, 32, 33, 35), while 4 trials did not report the proportion of it (24, 25, 26, 34). Four trials used background fructose 3% (3.2 to 18g) of total energy in the control groups, but put total fructose into consideration in the fruct

49、ose group (27, 28, 30-31). Only o ne trial reported less than 20g (4.3 % of total energy) fructose was consumed among basal diet (36). The baseline values were not provided in 5 trials (19, 22, 23). The median fructose dose in the available trials in cluded in our meta-a nalysis was 79.25 g/d (ran g

50、e: 30-182 g/d), and the durati on varied from 2 to 26 weeks.156157158159160161162163164165166167168169170171172173174175176177The quality scores of each study ranged from 6 to 9. Fifteen trials were classified as high quality (MQS 8), and 8 trials were of low quality (17, 19, 26, 30, 32-35). Only th

51、ree trials were blinded, one single-blinded (34) and 2 double-blinded (29, 35). Eight trials (19, 21, 24, 26-30) received in dustry fun di ng. Three studies with four trials (15, 16, 22) did not report any information about financial conflicts of interest. Effect of fructose on cholesterolTotal chol

52、esterol. Twenty-two trials (16-34, 36) reported the value of TC, and the pooled estimate was 2.47 mg/dL (95% Cl: -3.04, 7.98; P = 0.38) without statistically heterogeneity (heterogeneity ChF = 28.14, I2= 25%, P = 0.14) ( Fig. 1). The residual sources of heteroge neity were inv estigated by meta-regr

53、essi on models. Uni variate meta-regressi on showed that the fructose dose was positively related to TC, eve n after adjusted for study duration and health status (regression coefficient = 0.18; 95% CI:1781791801811821831841851861871881891901911921931941951961971981990.06, 0.31, P = 0.008) (Table 2)

54、. The dose-response relation between fructose consumption and TC largely explained the residual heterogeneity of the effect. Subsequently,we stratified fructose dose 60 to 100, and 100 as moderate, high, and very high, respectively, according to Candian Diabetes Association and refere nee ran ges fo

55、r fructose (9, 37, 38). Fructose could sig nifica ntly in crease TC by 12.97 mg/dL (95%CI: 4.66, 21.29; P = 0.002) whe n fructose in takes were 100g/d but had no effect on TC if fructose was given lower than 100g. Predefined subgroup an alyses were con ducted by study characteristicsSuppleme ntal Ta

56、ble 1). Sen sitivity analyses according to possible correlation coefficients (0.25 and 0.75) and systematically removal of each in dividual trial did not alter the overall an alysis and an alyses stratified by dose.LDL cholesterol. The mean change for LDL cholesterol in nineteen trials (15, 16, 18,

57、20, 22, 23, 25-35) was 3.76 mg/dL (95% CI: -1.07, 8.6; P = 0.13) without statistically heterogeneity (heterogeneity Chi2 = 19.85, I2 = 9%, P = 0.34) ( Fig. 2). The residual sources of heteroge neity were inv estigated by meta-regressi on models. Uni variate meta-regressi on showed that the fructose dose was positively related to LDL-C, eve n after adjusted for comparators, study duratio n and health statugegressi on coefficie nt =0.15; 95% CI: 0.03, 0.28, P = 0.02) (Table 2). The dose-response re