Prognosis of Inappropriate Left Ventricular Mass in Hypertension
The MAVI Study
To evaluate the prognostic impact of left ventricular (LV) mass exceeding individual needs to compensate hemodynamic load, the percentage of excess of echocardiographic LV mass in relation to individual ideal value predicted by gender, stroke work, and height (in meter2.7) from a reference population was assessed in 1019 white hypertensives (627 women [24% obese] and 392 men [17% obese, P<0.02 versus women]) without prevalent cardiovascular disease or type 1 diabetes, from the Italian multicenter, prospective study MAVI. Low LV mass (<73% of predicted) was found in 36 patients (3.5%), 661 had appropriate LV mass, and 322 (37%) had inappropriate LV mass. During follow-up (35±11 months), 52 fatal or nonfatal primary cardiovascular events occurred. Age, systolic blood pressure, and LV mass as a percentage of the predicted value were significant predictors of cardiovascular events (all P<0.01), independently of gender, glycemia, antihypertensive treatments, and body mass index, even in subgroups with or without LV hypertrophy. Survival analysis showed that cardiovascular risk increased stepwise from the lowest to the highest quintile of LV mass as a percentage of predicted value (P<0.01). The excess LV mass showed incremental prognostic value compared with assessment of traditional LV mass (P<0.01). Thus, inappropriate LV mass predicts a risk of cardiovascular events, independently of risk factors, and remains a significant predictor of risk either in the presence or in the absence of traditionally defined LV hypertrophy.
Left ventricular (LV) hypertrophy develops as the consequence of increase in LV mass (LVM) secondary to chronic overload. A LV anatomical adaptation that balances cardiac load is, therefore, compensatory. However, at least in arterial hypertension, a number of patients exhibit levels of LVM that exceed the need to sustain cardiac workload, a condition that has been defined as inappropriate LVM. The condition of inappropriate LVM is associated with high-risk phenotype1–4⇓⇓⇓ and is, at least in part, independent of the presence of traditionally defined LV hypertrophy.4 At the present, one study has demonstrated that in a selected group of hypertensive patients, the presence of inappropriate LVM increases the risk of adverse cardiovascular events.5 There is no confirmation of this finding on an epidemiological basis, nor is it clear whether the evaluation of LVM in terms of appropriateness to cardiac load offers any advantage in the modeling of cardiovascular risk over the more traditional assessment of clear-cut LV hypertrophy.
Accordingly, the present study has been designed to identify the physiological correlates and prognostic impact of inappropriate LVM in a large cohort of hypertensive patients and to assess whether it offers any incremental value in the prediction of risk compared with that obtained with more traditional methods.
The MAVI study is a multicenter study endorsed by the Italian Association of Hospital Cardiologists (ANMCO) to assess the prognostic value of echocardiographic LVM in Italian hypertensive patients.6 Admission criteria included age ≥50, absence of significant valvular disease,7 blood pressure (BP) ≥140/90 mm Hg or antihypertensive treatment, and no prevalent cardiovascular disease. Detailed information has been previously reported.6,8,9⇓⇓ Of the initially eligible participants, 627 (38%) were excluded because of the suboptimal technical quality of M-mode echocardiograms. Thus, the remaining 1019 participants (627 women [24% obese by National Institutes of Health criteria10] and 392 men [17% obese, P<0.02 versus women]) were considered for this study.
Outcome events included fatal or nonfatal myocardial infarction, sudden death, fatal or nonfatal stroke, other cardiovascular deaths, severe heart failure requiring hospitalization, severe renal failure requiring dialysis, documented angina, transient ischemic attack, and peripheral occlusive arterial disease verified by angiography.6,8,9⇓⇓
M-mode echocardiograms (2D targeted) were carried out in each hospital, recorded on videotapes, and sent to the Echo-Reading Center.6 LVM was calculated by a standard formula11,12⇓ and normalized for both body surface area (BSA) and height (meter2.7).13 LV hypertrophy was defined as LVM index >47 g/m2.7 in women and >50 g/m2.7 in men or as LVM index >125 g/m2 of BSA.14
LV end-diastolic and end-systolic volumes were calculated using Teichholz’s formula.15 Stroke volume was generated (mL/beat) and stroke work (SW in gram-meters/beat [g-m/beat]) was computed16 as follows:
LV systolic function was estimated as systolic shortening measured at the endocardial and midwall levels.17
Individual LVM was estimated using an equation previously developed in a separate reference population of 393 normal-weight, normotensive adults, aged 18 to 85 years16:
where male=1 and female=2. Observed LVM (oLVM) was divided by predicted LVM (pLVM) and was expressed as a percentage (oLVM/pLVM). With this method every individual served as a reference for him/herself.
For convenience, oLVM/pLVM was categorized using the 5th and the 95th percentiles of the distribution in the normotensive, normal-weight reference adult population.16,18⇓ Inappropriate LVM was defined as an excess of >28% from the predicted value (ie, oLVM/pLVM >128%) and low LVM as a decrease of >27% from the predicted value (ie, oLVM/pLVM <73%).
Standard laboratory tests included a primary work-up for hypertension, performed locally by standard methods. Diabetes was diagnosed by fasting venous plasma glucose ≥140 mg/dL or current treatment.19
Data were analyzed using SPSS 10.1 software (SPSS Inc) and expressed as mean±SD. Descriptive statistics were obtained using χ2 distributions (with Monte Carlo method for computation of exact 2-tailed α-value, when appropriate) and one-factor analysis of variance and the Ryan-Einot-Gabriel-Welsch (REGW) F test as post hoc evaluation, when needed. Participants were categorized into clusters according to either the presence of low, appropriate, or inappropriate LVM or the presence of incident cardiovascular events. Two-factor analysis of covariance (ANCOVA) was used to compare categories, accounting for significant confounders, identified in descriptive statistics. Log cumulative hazard functions were computed by Cox proportional hazards analysis, and survival curves were obtained. Cox-Snell residuals (individual cumulative hazard function estimate) were generated, including the effects of LVM-index and midwall shortening, and compared among patients with low, appropriate, or inappropriate LVM. The null hypothesis was rejected at a 2-tailed α of ≤0.05.
Comparison Between the Study Population and Ineligible Patients
As expected, the 627 patients with technically inadequate echocardiographic studies were slightly heavier (73.4±11.5 versus 71.9±11.7 kg, P<0.02) and older (61.3±7.2 versus 60.3±7.0 years, P<0.001) and more likely to be smokers (P<0.01) than individuals forming the study population, but no difference was found for body mass index or systolic and diastolic blood pressure. Gender distribution and the prevalence of diabetes were similar, and no difference was found in the survival function by Kaplan-Maier log-rank test (log-rank=0.37).
Characteristics of the Study Population
Age was comparable among women and men with appropriate (60.39±6.91 and 59.9±7.22 years, respectively) or inappropriate LVM (61.10±6.61 and 60.12±6.91 years), whereas participants with low LVM were younger (58.48±7.23 and 55.72±6.68 years, respectively; all post hoc, P<0.01).
The proportion of women and men was comparable in the groups with low (3.3% and 3.8%, respectively), appropriate (64.2% and 66.1%), or inappropriate LVM (32.5% and 30.1%). Among the 1019 studied patients, 135 had never been treated for arterial hypertension (13%), whereas the other 884 (87%) had been receiving some antihypertensive medications. Among them, 691 (68%) were regularly treated, whereas in 193, treatment was occasional (ie, several cycles of therapy over weeks or months, with an interval of months or years off drug). At the time of enrollment, 867 patients were on medications, whereas 17 individuals were off treatment (spontaneous withdrawal). The proportion of currently treated patients slightly increased from those with low (81%) to those with appropriate (84%) or inappropriate LVM (89%, P<0.07). Patients never treated had BP (157±18/95±9 mm Hg) similar to patients occasionally treated (157±17/93±9 mm Hg) and higher than patients on stable therapy (153±18/91±9 mm Hg, P<0.001). In contrast, oLVM/pLVM was lower in patients never treated (112±24%) than in patients on stable (119±31%) or occasional therapy (118±34; P<0.05), whereas LVM index was not different among subgroups. There was no difference in the proportion of currently treated women and men in all LVM strata. Details about the type of antihypertensive treatment have been previously reported.9
There was no difference in the proportion of prevalent diabetes and current or former smoking habit among the 3 subgroups in relation to the appropriateness of LVM in both genders.
Over a mean follow-up of 35±11 months, 53 patients (1 of 36 with low, 27 of 661 with appropriate, and 25 of 322 with inappropriate LVM) suffered at least one adverse fatal (n=9) or nonfatal cardiovascular event.
After controlling for age and antihypertensive treatment, systolic and pulse pressure, but not diastolic BP, were higher in the group with incident events, in either gender (Table 1). No significant differences were detected for plasma cholesterol, high-density lipoprotein (HDL)-cholesterol, triglycerides, and uric acid. Plasma creatinine was slightly higher in the subgroup with incident events. Plasma glucose was also higher in men and women with incident cardiovascular events than in event-free individuals, a difference that was fully confirmed also when BMI was considered as an additional covariate in the ANCOVA model (all probability values in Table 1).
According to the differences displayed in Table 1, the Cox proportional hazard model was generated including age, gender, systolic BP, plasma glucose, previous and present antihypertensive treatments, and oLVM/pLVM. Table 2 shows that age, systolic BP, and oLVM/pLVM were independent predictors of incident fatal and nonfatal cardiovascular events. Figure 1 shows the event-free survival curves in patients with low, appropriate, or inappropriate LVM, after adjusting for all covariates used in the proportional hazard model.
Survival curves were also obtained in quintiles of oLVM/pLVM from the entire study population and are displayed in Figure 2. As can be seen, a substantial increase in the probability of adverse events occurred at the fourth quintile of the distribution of oLVM/pLVM, corresponding to a cutoff value of 121%.
Relation With Traditionally Defined LV Hypertrophy
Among 483 patients with clear-cut height-based LV hypertrophy, 276 (57%) had inappropriate LVM, whereas among those without LV hypertrophy, 46 had inappropriate LVM. Proportional hazard analysis was repeated in patients without or with LV hypertrophy. In both strata, oLVM/pLVM demonstrated an additional predictive value. The probability of cardiovascular events was related to oLVM/pLVM, independently of other covariates, in patients with LV hypertrophy (P=0.05) and, especially, in those without clear-cut LV hypertrophy (P<0.03, Figure 3).
Cumulative hazard functions were individually computed (Cox-Snell residuals) from a Cox model including the covariates listed in Table 2, but substituting oLVM/pLVM with LVM index (in grams per meter2) and adding midwall shortening as another proved independent marker of cardiovascular risk.20 The cumulative hazard function was similar in patients with low (0.03±0.02) or appropriate LVM (0.04±0.04), but substantially higher in those with inappropriate LVM (0.07±0.06, P<0.0001). Similar results were also obtained when using the LVM index in grams per meter2.7 in the Cox model.
There is physiological justification for generating a complex calculation in the attempt to discriminate the LV growth needed to sustain hemodynamic load from the LV growth that is no longer mediated by recognizable hemodynamic stimuli.16,21⇓ This approach is a cultural derivative of the procedure of adjusting LVM for stroke work, used years ago as a measure of LV performance.22,23⇓ The practical advantage of calculating the ratio between observed LVM and predicted LVM is that each subject serves as his or her own reference, so there is no need to use relatively arbitrary partition values, often obtained in different populations, to define excessive levels of LVM.
An adverse impact of inappropriate LVM on cardiovascular risk profile has been previously observed in a number of cross-sectional1–4⇓⇓⇓ and one longitudinal5 studies, undertaken in different hypertensive and normotensive populations. The present study provides a number of new insights into the issue.
Probability of Adverse Events in the MAVI Cohort
This is the first large-scale multicenter study demonstrating the prognostically independent relevance of inappropriate LVM. Compared with the only available single-center study,5 a more complete proportional hazard model was used, including a larger number of covariates found to be associated with cardiovascular morbidity in descriptive statistics. The prognostic impact of LVM expressed as a percentage (%) of individually predicted reference value was comparable to the one reported in the previous study.5 The deviation of observed LVM from the individual reference value permitted discrimination of cardiovascular risk also in subgroups of patients with or without traditionally defined LV hypertrophy. It is relevant that in the group of patients without LV hypertrophy, those with inappropriate LVM showed high cardiovascular risk of adverse events despite the normal levels of LVM. According to the traditional definition of target organ damage (ie, LV hypertrophy), cardiovascular risk would be underestimated in this group. Consistent with this finding, the individual cumulative hazard function resulting from a risk assessment including the effect of LVM index and midwall shortening was nearly 2-fold higher in patients with inappropriately high LVM, demonstrating a substantial increase in prognostic information over the one provided by LVM index and systolic function measured at the midwall level.
Figure 2 shows that cardiovascular risk progressively increases with higher levels of the ratio of observed LVM to predicted LVM. Interestingly, the risk began to rise after 100% (which corresponds to the theoretic individual normality, ie, equality between observed and predicted LVM) and was sharply higher when LVM exceeded the value of 120%. Thus, the near identity between observed and predicted LVM can be considered as a compensatory adaptation, independent of the magnitude of absolute LVM values, whereas every deviation from 100% may be considered as supraphysiologic, or frankly pathologic, depending on the magnitude of the excess.
Another new, original aspect of this study is that the MAVI is a cohort with a relatively low prevalence of obesity, at least in part because of the selection of patients requiring an optimal M-mode echocardiogram. Although it is reported that oLVM/pLVM and the proportion of inappropriate LVM increase with obesity, the present study demonstrates that the association of high-risk phenotype with inappropriate LVM is also largely independent of obesity or body mass index. This was actually expected because, if the use of height in the equation predicting LVM tends to highlight the effect of obesity, stroke work tends to balance that effect because it is substantially higher in obese individuals, yielding therefore increased predicted values.21
There might be 2 limitations to this study that need to be highlighted and discussed. This cohort is selected on the basis of good quality M-mode echocardiogram, excluding more than 35% of the initially eligible patients from the follow-up analysis. This selection has excluded high-risk patients and also explains the low prevalence of obesity in this cohort. Thus, the outcomes could be more striking if the excluded high-risk patients were also considered. However, this limitation also carries an advantage, ie, it allowed the possibility of evaluation without a strong effect from obesity, as reported in previous studies.
Most patients were treated, and covariance models cannot offset completely the biological effect of therapy. This is a common problem that markedly affects cross-sectional evaluations of raw parameters that are targets of treatment (ie, LVM). In this case, however, LVM is evaluated in relation to loading conditions in a context in which the reference is the theoretical value of 1:1 of observed:predicted. The effect of treatment is less important than when evaluating the absolute value, exposed to changes in relation to the therapy-induced changes in hemodynamic load.
The gender-specific LVM that is in excess of theoretical values appropriate for individual stroke work and body size increases progressively the risk of cardiovascular adverse events in a large multicenter white cohort from Italy, independently of recognized cardiovascular risk factors, age, and blood pressure. Evaluation of LVM as a percentage of the individual reference value can contribute to improvement in our ability to identify patients with high cardiovascular risk. Because this approach also allows recognition of high-risk hypertensive patients who do not have traditionally defined LV hypertrophy, as well as normotensive subjects who might be at high cardiovascular risk, future research should investigate the meaning of inappropriate LVM even in those subsets of population.
E. Giovannini (chairman), G.C. Carini, A. Circo, E.V. Dovellini, M. Lombardo, P. Solinas, P. Verdecchia
Scientific and Organizing Secretariat
S. Ghione, M. Gorini, D. Lucci, A.P. Maggioni
Echocardiogram and ECG Central Reading
C. Borgioni, A. Ciucci, G. Gozzellino, A. Milletich
Participanting Clinical Centers
Avola (E. Mossuti, G. Canonico, G. Romano), Bari Policlinico (I. De Luca, N. Ciriello), Bazzano (A. Baldini, G. Castelli), Belluno (G. Catania, L. Tarantini), Bologna Osp. S. Orsola-M. Malpighi (E. Ambrosioni, F. Marchetta), Brindisi Osp. Di Summa (G. Ignone, D. Zuffianò, A. Storelli), Cagliari Osp. Binaghi (G. Isaia, E. Muscas), Camerino (R. Amici, B. Coderoni), Capua (E. Mercaldo, R. Esposito), Casale Monferrato (M. Ivaldi, G. Gozzelino, A. Capilli), Casarano (G. Pettinati, M. Ieva, A. Marzo), Caserta (G. Corsini, S. Romano, A. Martone), Catanzaro Policlinico (F. Perticone, C. Cosco), Chieti Villa Pini D’Abruzzo (C. Ciglia, P. Di Giovanni), Citta’ Della Pieve (G. Benemio, G. Schillaci, N. Sacchi), Cosenza INRCA (E. Feraco, A.M. Nicoletti), Desio (V. Baldini, M. Cristofari), Firenze Osp. Careggi (P.F. Fazzini, D. Antoniucci, E.V. Dovellini, E. Taddeucci), Firenze Osp. Camerata (F. Marchi, A. Buzzigoli), Foligno (M. Massi Benedetti, C. Pagnotta, R. Liberati), Gallarate (R. Canziani, G. Cozzi, M. Alberio), Gioia Del Colle (F. Barba), Gorizia (A. Fontanelli, R. Marini, L. Scarpino), Grottaglie (V. Portulano, G. Sportelli), Matera (L. Veglia, T. Scandiffio), Mesoraca (F. Schipani, C. Tangari), Messina Osp. Papardo (L. Cavallaro, G. Sergi, S. Mangano), Messina Policlinico Universitario (S. Coglitore, C. De Gregorio, D. Cento), Milano Osp. Niguarda (A. Pezzano, M. Lombardo), Monfalcone (T. Morgera, G. Zilio, D. Chersevani), Napoli Osp. Cardarelli (L. D’Aniello, F. D’Isanto), Oliveto Citra (G. D’angelo, V. Iuliano, P. Bottiglieri), Palermo Osp. Ingrassia (P. Di Pasquale, F. Clemenza, S. Cannizzaro, G. Caramanno), Perugia (C. Porcellati, P. Verdecchia, A. Ciucci, C. Borgioni), Pieve Di Cadore (J. Dalle Mule, M. Mazzella), Polistena (R.M. Polimeni, F. Catananti, F. Terranova), Pontedera (G. Squarcini, S. Giaconi), Pordenone (G.B. Cignacco, G. Zanata, D. Pavan), Potenza (A. Lopizzo, M. Chiaffitelli, M. Faruolo), Prato (A. Petrella, M. Paoletti), Roma CTO (M. Uguccioni, D. Mocini), Roma Osp. Forlanini (A. Majid Tamiz, A. Avallone), Roma Osp. San Camillo II Divisione (E. Giovannini, A. Chiantera, F. De Santis), Roma Osp. San Camillo Servizio (P. Tanzi, L. Boccardi, D. Colecchia), Rovigo (P. Zonzin, A. Bortolazzi, S. Aggio), San Giovanni Rotondo (R. Fanelli, A. Russo, A. De Vita), San Pietro Vernotico (L. Vergallo, S. Pede, A. Renna), Sarzana (G. Filorizzo, D. Bertoli, M. Corradeghini), Scorrano (E. De Lorenzi, A. Bergamo, A. Colizzi), Sesto San Giovanni (R. Melloni, E. Chiocca), Sondalo (G. Occhi, G. Frisullo), Sondrio (S. Giustiniani, M. Marieni, M.L. Ghirimoldi), Soveria Mannelli (G. Bellieni, A. Marotta, A. Andricciola), Termoli (D. Staniscia, T. Alfieri), Udine (P. Fioretti, L. Pilotto), Vasto (G. Di Marco, G. Levantesi, L. Cavasinni), Viareggio (A. Pesola, G. Marracci, M. Pardini), Viterbo (R. Guerra, A. Achilli, S. De Spirito, R.Castellani).
The MAVI Study was endorsed by the Italian Association of Hospital Cardiology (ANMCO) and partially supported by Pfizer Italia spa.
- Received June 27, 2002.
- Revision received July 17, 2002.
- Accepted August 13, 2002.
Palmieri V, de Simone G, Roman MJ, Schwartz JE, Pickering TG, Devereux RB. Ambulatory blood pressure and metabolic abnormalities in hypertensive subjects with inappropriately high LV mass [published erratum appears in Hypertension. 2000;36:147]. Hypertension. 1999; 34: 1032–1040.
Palmieri V, Wachtell K, Gerdts E, Bella JN, Papademetriou V, Tuxen C, Nieminen MS, Dahlof B, de Simone G, Devereux RB. Left ventricular function and hemodynamic features of inappropriate left ventricular hypertrophy in patients with systemic hypertension: the LIFE study. Am Heart J. 2001; 141: 784–791.
US-DHHS. Cardiac valvulopathy associated with exposure to fenfluramine or dexfenfluramine. US Department of Health and Human Services Interim Public Health Recommendations, November 1997. Morb Mortal Wkly Rep. 1997; 14: 1061–1066.
Sahn DJ, DeMaria A, Kisslo J, Weyman A. Recommendations regarding quantitation in M-mode echocardiography: results of a survey of echocardiographic measurements. Circulation. 1978; 58: 1072–1083.
de Simone G, Devereux RB, Kimball TR, Mureddu GF, Roman MJ, Contaldo F, Daniels SR. Interaction between body size and cardiac workload: influence on left ventricular mass during body growth and adulthood. Hypertension. 1998; 31: 1077–1082.
de Simone G, Devereux RB, Roman MJ, Ganau A, Saba PS, Alderman MH, Laragh JH. Assessment of left ventricular function by the midwall fractional shortening/end-systolic stress relation in human hypertension [published erratum appears in J Am Coll Cardiol. 1994;24:844]. J Am Coll Cardiol. 1994; 23: 1444–1451.
de Simone G, Roman MJ, Daniels SR, Mureddu G, Kimball TR, Greco R, Devereux RB. Age-related changes in total arterial capacitance from birth to maturity in a normotensive population. Hypertension. 1997; 29: 1213–1217.
Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. National Diabetes Data Group. Diabetes. 1979; 28: 1039–1057.
de Simone G, Devereux RB, Koren MJ, Mensah GA, Casale PN, Laragh JH. Midwall left ventricular mechanics. An independent predictor of cardiovascular risk in arterial hypertension. Circulation. 1996; 93: 259–265.
de Simone G, Pasanisi F, Contaldo F. Link of nonhemodynamic factors to hemodynamic determinants of left ventricular hypertrophy. Hypertension. 2001; 38: 13–18.