Comparison of Calculated LDL-Cholesterol Using Various Formulae with Directly Measured LDL-Cholesterol: A Retrospective Study

Background: Among the various risk factors for developing dyslipidemia and Cardiovascular Disease (CVD), elevated Low-Density Lipoprotein Cholesterol (LDL-C) has been recognized as an independent and widely accepted risk factor for CVD. Therefore the objective of the study was to calculate LDL-C using various formulae and to compare the various formulae to calculate LDL-C with direct LDL-C measurement. Materials and methods: It was a retrospective observational study and study was conducted for a period of 3 Months. During this period, 400 consecutive serum samples were included, out of which 37 had TG value more than 400 mg/dl and were excluded from the study. Finally 363 samples data were included for analysis and data were obtained from the investigation ledger of clinical Biochemistry section and same sample Triglyceride, Total Cholesterol and High Density Lipoprotein Cholesterol was used for calculating LDL-C using various formulae. Results: A total of 400 lipid profile reports were collected, of these, 37 reports had triglyceride level ≥ 400 mg/dl and hence 363 samples were included. There is no statistical difference between the direct LDL-C (131.64± 29.34mg/dl) and LDL-C level calculated by Friedewald formula (133.85±35.97mg/dl, p>0.05). There is a highly significant statistical difference between the direct LDL-C and the LDL-C level calculated using all other formulae (p=0.0001) Cordova and Cordova formula (88.06 ± 28.40mg/dl), Vujovic formula (80.44 ± 15.52mg/dl), Ahmadi formula (186.34 ± 63.26mg/dl), Anandaraja formula (92.46 ± 32.37mg/dl), Puavillai formula (88.54 ± 32.78mg/dl) and Hattori formula (77.45 ± 30.31mg/dl). Conclusion: The study established Friedewald formula as the most suitable method for calculating LDL-C in the absence of direct LDL-C measurement facilities. A large sample size and multicentric study is warranted to confirm which formula is most suitable for measuring the LDL-C in the absence of direct LDL-C measurement.


Introduction
Cardiovascular disease (CVD) is a group of disease of heart and blood vessels and it is one of the most common causes of morbidity and mortality worldwide. To identify the risk factors of CVD, lipid profile is done routinely as a screening test. It includes the estimation of Triglycerides (TG), Total Cholesterol (TC), High-density Lipoprotein Cholesterol (HDL-C), Low-density Lipoprotein Cholesterol (LDL-C) and Very low density Lipoprotein Cholesterol (VLDL-C). [1] Among the various risk factors for developing dyslipidemia and CVD, elevated LDL-C has been recognized as an independent and widely accepted risk factor for CVD. [1] The National Cholesterol Education Programme (NCEP) Adult Treatment Panel III (ATP III) recommends a goal of maintaining serum LDL-C concentration < 100 mg/dl as optimal. Depending on the value of serum LDL-C, adults are grouped into various risk of developing heart disease if: LDL-C value is 100-129 mg/dl -Desirable Risk, LDL-C value is 130-159 mg/dL -Borderline high-risk and LDL-C value is ≥160 mg/dL considered as High risk. [2] In 1983 homogenous assays was developed for direct measurement of LDL-C. [3] In India this method is not routinely practiced in most of the laboratories because direct LDL-C estimation is expensive and time consuming. Hence, majority of the clinical laboratories measure LDL-C concentration in serum using Friedewalds formula by taking the concentration of TC, HDL-C and TG, with the help of formula LDL-C=TC-( HDL-C+TG/5). [4] There are several shortcomings of Friedewalds formula, it cannot be used for LDL-C calculation when the patient is not in fasting, if serum TG>400 mg/dL and in patients of hyperlipoproteinemia. [5] Despite these limitations, Friedewalds formula is still the most commonly used formula in clinical laboratories for measuring LDL-C concentration. [6] Several other formulae have been proposed over the years for calculation of LDL-C. These are Cordova and Cordova [7] et al., Vujovic [8] et al., Ahmadi [9] et al., Anandaraja [10] et al., Puavillai [11] et al., and Hattori 12] et al., Similar to Friedewald's formula, these formulae also use TG, TC and HDL-C to calculate LDL-C. These formulae are not without the limitation. By considering this ambiguity in reporting, the present study was undertaken with the objectives to calculate LDL-C using various formulae and to compare with direct LDL-C assay, so as to obtain an acceptable formula at our setting in replacement for the direct measurement hence making lipid profile estimation cost effective.

Material & Method
This was a retrospective observational study and study was initiated after obtaining approval from the Institutional Ethics Committee of Adichunchanagiri Institute of Medical Sciences (AIMS), B G Nagara. The fasting serum sample received at Biochemistry section of clinical Laboratory for estimating the lipid profile was analyzed using various enzymatic methods.
The data required for the present study was obtained from the investigation ledger of Biochemistry section of clinical Laboratory and same sample TG, TC and HDL-C was used for calculating LDL-C using various formulae.
The present study was carried out for a period of 3 Months. During this period a total 400 consecutive serum samples that were in the age group between 18-70 years, received in the clinical laboratory for estimating lipid profile were considered. Out of 400 samples, 37 had TG value more than 400 mg/dl and were excluded from the study. Finally 363 samples data were included for analysis.
Lipid profile was analyzed using standard methods such as TC was analyzed by cholesterol oxidase peroxidase method. TG was estimated by glycerol peroxidaseperoxidase (GPO-PAP) method. HDL-C was analyzed by PEG -cholesterol esterase (direct enzymatic) method. The LDL-C assay was analyzed by cholesterol esterasecholesterol oxidase (homogeneous) method (Direct LDL-C).
LDL-C was also calculated by using various formulae by considering the values of TC, serum TG, HDL-C data from the investigation ledger. The factor [TG] / 5 is an estimate of VLDL-C concentration, and is based on average ratio of TG to cholesterol in VLDL-C.

Statistical Analysis
Data were entered into Microsoft Excel sheet and analyzed using SPSSv15 software. Results are represented as mean ± standard deviation. The statistical significance was evaluated at 95% confidence level and probability value (p) of < 0.05 was considered as statistically significant.

Results
Lipid profile data including direct LDL-C were obtained from the investigation ledger of Biochemistry section of clinical Laboratory retrospectively from past 3 months. A total of 400 lipid profile reports were included. Of these, 37 reports had triglyceride level ≥ 400 mg/dl, and hence were rejected before final analysis.
The mean age distribution of the subject was 50.38±11.92 years with male as 50.23±11.44 years and female as 50.67±12.28 years (p value >0.05). Figure 1 shows the distribution of male and female study population and the male to female ratio was 1.8:1.   Table 2 shows distribution of study Population having the normal and abnormal values of lipid profile. Majority of the study population were having total cholesterol (74.38%) and HDL-C (85.39%) in reference range. However more than half of the study population was having higher triglycerides (55.37%) and another 78.78% of them were having LDL-C above reference range.  Table 3 & figure 2 shows the comparison of the estimated LDL-C (Direct LDL-C) levels with the calculated LDL-C values by various formulas. There is no statistical difference between the direct LDL-C and LDL-C levels calculated by Friedewald formula (p>0.05). There is a highly significant statistical difference between the direct LDL-C and the LDL-C levels calculated using all other formulae (p=0.0001) except Friedewald formula. The LDL-C levels are overestimated when Ahmadi formula is applied whereas the values are underestimated by using the other formulae. A similar significance was also observed when comparison was done between the direct LDL-C and calculated LDL-C levels at different LDL-C ranges as shown in table 4. Mean value of lipid profile and calculated LDL-C using various formulae did not show gender wise difference and it was found to be statistically insignificant as shown in table 5.

Discussion
The study compared several formulae including Friedewald formula with direct LDL-C measurement. The measurement of LDL-C level is considered as a predictor for assessing the risk of various cardiac diseases and for treatment of dyslipidemias. Even though various formulae have been developed to calculate the LDL-C, most of the laboratories use Friedewald formula as the choice of formula to calculate LDL-C but many studies have shown its limitation and some have shown that other equations perform better for certain groups of populations.
The important finding of present study is that LDL-C level calculated using Friedewald formula did not show statistical significant difference with the directly measured LDL-C. But the LDL-C levels are overestimated when Ahmadi formula is applied whereas the values are underestimated by using the other formulae except Friedewald formula as seen in Table 3 & Graph 2.
The present study results are in concurrence with the study done by Niranjan et al. In this study on comparison of Anandaraja Formula and Friedewald Formula with Direct LDL-C measurement, they concluded that Friedewald formula showed a strong correlation (0.876) in comparison with Anandaraja formula (0.844). [13] In another study by Nishtha Wadhwa [14] et al., it was concluded that in Indian Population, Vujovic formula appears to be more accurate than any other formula. Whereas in another study conducted by Mugdha Dilip Garule [15] et al, it was reported that, Puavillai formula is the most accurate formula to calculate LDL-C at TG levels up to 150 mg/dL and also at all TG levels studied whereas Friedewald's formula is the best at TG levels between 151 to 199 mg/dL and Anandaraja formula at TG 200 to 399 mg/dL. However in the present when the comparison was done even at different levels of LDL-C (ie optimal level, desirable level, borderline and high-risk level), calculated LDL-C using Freidewald formula did not show significant difference whereas the other formulae showed significant difference with Ahmadi formula overestimating and others formulae the underestimating the LDL-C values at all levels of LDL-C as shown in table 4.
Farideh Razi [16] et al in a study suggested calculation of LDL-C based on Friedwald and Chen formula can be a good alternative for direct measurement especially in regions with limited resources. This study results are in concurrence with our study results. In our study we found that compared to other formulas, LDL-C calculated using Friedewald formula correlates better with direct LDL-C.

Conclusions
To conclude, the study established Friedewald formula as the most suitable method for calculating LDL-C in the absence of direct LDL-C measurement facilities. The limitation of the present study is small ample size and comparison was not done at different levels of triglyceride including when triglyceride levels were greater than ≥400 mg/dl. A large sample size and multicentric study is warranted to confirm which formula is most suitable for measuring the LDL-C in the absence of direct LDL-C measurement.
Further the authors propose to continue future studies comparing the usefulness of these formulae in calculating LDL-C when triglycerides are ≥400mg/dl which is considered as a limitation when Friedewald formula is applied.