Retrospective #476: Martin/Hopkins, “a better LDL calculation method”

In mid-2018 I noticed that Quest Laboratories had changed their method of calculating low-density lipoprotein (LDL-C) on my lab report. For more than 40 years the method used to calculate it was the Friedewald formula (LDLC = TC – HDL – TG/5), where TG/5 (TG<400mg/dl) was a surrogate for VLDL cholesterol. The method Quest uses now is the Martin/Hopkins method. Quest says that this “novel” method is “more accurate than the usual method.”

Of course, the most accurate method to measure LDL-C is a DIRECT assay, not a calculation. Dr. Michael Eades explains this on his website and I discuss it here. It requires another test – a special test which is expensive and generally not done unless you request it and your doctor and insurer agree.

I’m writing about this change because, based on what Dr. Eades wrote, and from my own affirming data, I assumed that Quest changed to the new method because the new method reported a lower LDL-C. To my surprise, while doing some research recently, I discovered that the new method actually shows a higher LDL-C.  I found the story here in MEDPAGE TODAY, under Cardiology > Dyslipidemia. The sub-title of the article is, “Friedewald lowballed lipids; Martin/Hopkins was on par with lab reference” [for patients with stable CVD].

“According to an analysis of the FOURIER trial, the median Martin/Hopkins LDL cholesterol level was 2mg/dl below the reference standard of preparative ultracentrifugation – a statistically similar result, whereas the Friedewald method underestimated LDL cholesterol by 4mg/dl (P<0.001)” (all emphases added by me). So, relative to Friedewald, Martin/Hopkins yielded an LDL-C that was closer to “reference” (assayed value) and 2mg/dl higher than Friedewald.

“While 22.9% of Martin/Hopkins LDL cholesterol values were at least 5mg/dl different from reference and 2.6% were off by more than 10mg/dl, these proportions were 40.1% and 13.3% with Friedewald estimation.” Noto Bene: “The difference between methods was more pronounced when triglyceride levels exceeded 150mg/dl.”

While the MEDPAGE TODAY article did not give the whole Martin/Hopkins formula, it did reveal that “the Martin/Hopkins method ‘uses the same standard lipid measurements of total and HDL (high-density lipoprotein) cholesterol and triglycerides as the Friedewald equation does, but it uses a personalized rather than a fixed conversion factor in calculating LDL cholesterol levels,’ Martin and colleagues noted.” This is “the nut” of it.

But, I wondered, does this apply equally to low triglycerides? If Quest’s use of a personal conversion factor usually results in an LDL-C higher (even if only 2 mg/dl), I find it intriguing that on the three lab reports I’ve had since Quest switched to the Martin/Hopkins method, my personal LDL-C values have all been lower than they were before. This is why I presumed, based on what Dr. Eades had said about a DIRECT measurement, that my own values would probably be lower, not higher, using the new method as it was intended.

For the 3 tests before the switch from Friedewald to Martin/Hopkins, my LDL-C values were 101, 114 and 100mg/dl. Then, using Martin/Hopkins, the Quest reports for my last 3 labs were 87, 79 and 83mg/dl. Of course, my labs were for different samples, but I have not changed my Way of Eating. So, then, I wondered, why are my LDLs lower?

Cholesterol	198	201	196	184	173	189
HDL	85	74	74	83	81	92
Chol/HDL ratio	2.3	2.7	2.6	2.2	2.1	2.1
LDL (calc.)	101	114	100	87*	79*	83*
non-HDL	113	127	122	101	92	97
Triglycerides (TG)	60	67	108**	56	53	56

* calculated by Martin/Hopkins method   ** first TG > 100mg/dl in 12 years

And here’s the answer: The JAMA article explains: Martin/Hopkins uses statistics for 3 stratifications of TGs from 100 to 399mg/dl, which resulted in higher LDLs. My lower LDLs are because my 3 most recent TGs have all been in the mid-50s. Martin/Hopkins does not account for low TGs. They’re “off the chart,” so to speak!