After looking at the observational studies regarding saturated fat (SFA) and coronary heart disease (CHD) two things stuck out: (1) SFA wasn’t associated with CHD; and (2) PUFA was sometimes inversely associated with CHD (which may depend somewhat on whether the Harvard School of Public Health is writing the meta-analysis).
One could dismiss the results by taking the approach of correlation ≠ causation, observational studies have confounding variables, etc . But the inverse association could also be real (not due to confounding variables) and then if so, what is the effect caused by? The observational studies look at PUFA, SFA, etc, but we don’t eat nutrients, we eat food. What does replacing PUFA with SFA or CHO translate to when we think about foods consumed in a population?
Vegetable oils are a major source of PUFA, but this inverse association between PUFA and CHD (assuming it’s real) is probably not due to vegetable oils. The clinical trials that tested replacing SFA with vegetable oils show that it takes a multifactorial diet (ODHS, STARS), a higher intake of trans fats and higher use of cardiotoxic drugs in the control group (FMHS) or a vitamin E deficient control diet (LA Vets) for there to be any benefit regarding CHD events/mortality (total mortality is another story). When these things are absent, or near absent (RCOT, MRC, SDHS, MCS, DART), there is no difference on average.
Another major source of PUFA could be nuts as they are very dense sources of PUFA by weight and have a relative high proportion of their calories come from PUFA. So it’s easy to see how even a ‘small’ quantity of nuts by weight can strongly influence PUFA intake. (Remember that high PUFA intake is 10% of total calories, which is 22g of PUFA in a 2,000 calorie diet)
PUFA per 100g
PUFA % of Energy
Calories per 100g
Observational studies have generally shown nuts to be fairly strongly inversely associated with CHD, often having one of the strongest inverse associations among all foods tested  . The strong inverse association between nuts and CHD may then be diluted by other things high in PUFA like vegetable oils, resulting in the weaker association between PUFA and CHD
Mente, et al (2009) 
High vs. low
Kelly & Sabate (2006) 
High vs. low
Kelly & Sabate (2006) 
Unit* per week
* Probably an ounce (28.3g)
Assuming this association is also not due to confounders, why are nuts inversely associated with CHD? On paper nuts don’t have much going for them. Compared to other whole foods they are low in protein and have the lowest nutrient density per calorie  , a really high omega 6:3 ratio and a very high energy density.
There are some proposed mechanisms for how nuts may be cardioprotective, which are outlined in these papers*  . One of the most commonly mentioned mechanism is that nuts (and seeds) are high in arginine, which is a precursor of nitric oxide, which promotes endothelial function.
However, the effects of supplemental arginine are often minor and/or inconsistent  and while nuts (and seeds) have a very high proportion of arginine relative to total amino acids (~13-14% vs. ~6-8% in other whole foods) they aren’t good sources of protein so total arginine is average (although the ratio may be more important). Also, many of the other mechanisms (such as inflammation) probably also apply to other whole foods. It could simply be that as most of time, most people eat nuts as a snack, which could mean displacing other snack foods**, which are often highly processed, very low in nutrients and may contain questionable ingredients
- Nuts may reduce the risk of CHD, as there’s certainly a strong inverse association in observational studies, but this may depend on what they replace
- The strong inverse association between nuts and CHD may explain the weaker inverse association between PUFA and CHD in observational studies
* The authors of these papers have been funded by industry
** This is similar to observational studies showing a benefit of whole grains probably because they replaced refined grains