From the Masters: Lumper? Or splitter?

I know a vascular medicine physician who’s a ‘splitter’. He’s never met a disease he can’t subdivide into multiple diseases. His patients don’t just have lipedema – they have ‘type 2 lipedema’. For those who may be unfamiliar with this classification system, the distribution of type 2 lipedema involves the area from the buttocks to the knees, with folds of fat typically deposited around the medial thigh. Four other types, based on the distribution of fatty tissue, have also been described. Because there are also four stages of this condition, his patients have things like ‘type 2, stage III lipedema’ (stage III involves nodular fat formation, especially in the thighs and knees). Five types, four stages – that’s 20 possible type/stage subgroups of lipedema – and he fits each of his patients neatly into one of these niches.

My patients? They just have lipedema.

Here’s the difference. He’s a splitter. I’m a ‘lumper’. He looks at the world and sees differences. I see similarities. I know I’m supposed to be a splitter. I am, after all, a medical specialist – someone who splits diseases into progressively smaller categories, steadfastly learning ‘more and more about less and less, until I eventually know everything about nothing.’ Conventional wisdom suggests that we lumpers ought to be big fat failures as medical subspecialists. Yet somehow a few of us succeed. Perhaps there’s a role for lumpers in a specialty like vascular medicine?

I don’t recommend wracking your brain over this question. You can’t use logic to decide whether to lump or split. I was taught in college that deduction is superior to induction for solving problems. Deduction starts with general points and evolves to reach specific conclusions: All vascular medicine physicians in Sioux Falls are good doctors. ‘M’ is a vascular medicine physician in Sioux Falls. Therefore, ‘M’ is a good doctor. This assessment, the result of deductive logic, certainly seems correct. Induction, on the other hand, starts with specific points and evolves to reach general conclusions: ‘J’ is a good vascular medicine physician. ‘J’ lives in Cleveland. Therefore, all good vascular medicine physicians live in Cleveland. This assessment, the result of inductive logic, somehow seems incorrect. If deduction (general to specific) is analogous to splitting, and induction (specific to general) is analogous to lumping, shouldn’t splitting be a superior, more logical approach to vascular medicine than lumping?

Hardly! You’ve just fallen for a textbook example of false analogy. Lumping and splitting are based upon the presence of similar (or dissimilar) characteristics and patterns. There’s no logic involved. Let’s split up the jelly beans, Heather. Red ones in this box. Blue ones over there. In vascular medicine, we likewise split things like disease into subcategories based upon easily discernible, often obvious distinguishing features, not logic. This may not always be the best strategy – splitting stuff endlessly into smaller and smaller categories is all fun and fuzzy puppies until someone accidently splits an atom and the whole thing blows up. Let’s explore the lumper/splitter quandary by looking at four areas within vascular medicine where aggressive splitting may not have helped as much as we hoped it would.

Large vessel vasculitis – splitting based on age

Takayasu arteritis and giant cell arteritis (GCA) are the two major entries in the category of large vessel vasculitis. Conventional wisdom considers them to be two different diseases. Are they? They have the same histology. The angiographic/anatomical appearance is nearly identical in both (you can argue that different arteries are clinically involved to a greater or lesser extent in one disease or the other, but we all have seen angiograms where you absolutely, positively can’t be certain which disease you’re looking at based solely on the radiographic appearance). Treatments, generally involving immunosuppression, are similar. There are plenty of other ways in which these two versions of vasculitis appear to be identical. So, if they are truly different, then how do we tell them apart?

Here’s how we do it in practice. You show me an angiogram and ask, ‘Does this look like Takayasu or GCA?’ I reply, ‘How old is the patient?’ You answer, ‘Thirty.’ I say, ‘Thirty? It’s obviously Takayasu.’ Then you say, ‘Oh, sorry – I was thinking of the wrong patient. This lady’s actually 68.’ And I say, with a completely straight face, ‘Sixty-eight? It’s obviously giant cell.’ This improbable conversation occurs because the main factor differentiating Takayasu from GCA is ‘the age of the patient’. Does this make sense? Not to me. Rather than designating two separate vasculitides that exclusively strike people at opposite ends of the age spectrum, why aren’t we considering this to be a single vasculitis that (1) affects anyone, and (2) presents differently in patients of different ages? After all, illnesses like chickenpox (along with mumps and many other childhood diseases) don’t affect children the same way they do adults; for example, kids with chickenpox typically have a mild, self-limited, benign presentation that keeps them home from school for a day or two, while adults may develop severe problems like pneumonia or encephalitis that can put them in the hospital. Although the adult and juvenile forms of chickenpox act like completely different afflictions – with different symptoms, different prognosis, and different complications – we don’t treat them like different diseases, and we don’t give them different names. Instead, we lump the pediatric and adult forms into a single disease and call them both ‘chickenpox’. If this is how we characterize chickenpox, why do we insist on splitting large vessel vasculitis into two separate diseases based primarily on the age of the patient? Why are we referring to ‘Takayasu’ and ‘giant cell’ arteritis instead of something more descriptive like ‘early-onset GCA’ and ‘late-onset GCA’? ¹

Fibromuscular dysplasia (FMD) and segmental arterial mediolysis (SAM) – splitting based on patient sex

We treat FMD and SAM as separate diseases, yet they’re nearly identical with regard to pathology, distribution of affected vessels, prognosis, complications, and many other features. Experts write papers where they claim in one part of a sentence that there are ‘distinct clinical differences between these two disorders’ only to concede in another part of the same sentence that ‘SAM lesions closely resemble … FMD.’ ² I have a friend who is one of the world’s leading authorities on FMD; he suggests: ‘It is difficult to differentiate FMD from SAM. It is unclear whether SAM is a distinct vascular abnormality or a variant of FMD.’ ³ I’m confused. Are FMD and SAM one disorder? Or two? And, if they are two distinct entities, then how do we tell them apart?

Some try to distinguish between FMD and SAM based upon differences in their respective clinical course. It’s been alleged, for example, that ‘SAM is simply FMD that dissects.’ I confess – I’ve occasionally used the presence of dissection to argue for a diagnosis of SAM as opposed to FMD, but we all know that’s not how clinicians usually differentiate between them in everyday practice. Most clinicians discriminate between the two based on the observation that: ‘Fibromuscular dysplasia is a … disease that predominately affects women.’ ⁴ Indeed, the female:male ratio is reportedly as high as 9:1. ⁵ Contrast this to the situation with SAM, where there’s a definite predominance of males; one recent study of patients with SAM contained 2.5 times as many males as females. ⁶

A cynic might argue that the biggest – and perhaps only – difference between FMD and SAM is the sex of the patient: if this is a single disease and it occurs in females, we call it FMD; in males, it’s SAM. Are we really looking at two different entities? Or is this a single entity that presents differently in females versus males, akin to the way chickenpox (see above) presents differently in kids as opposed to adults? I prefer to think of FMD and SAM as a single disease with different presentations: one in which women are more likely to exhibit the classic arterial beading associated with FMD, while men are more inclined to develop the dissections we typically attribute to SAM. I suspect we err by splitting FMD and SAM into two different diseases rather than treating them as one.

Atherosclerosis (ASO) – splitting based on regional vascular involvement

The pyramids were built with less effort than we’ve put into splitting ASO into various categories. We treat coronary artery disease, carotid artery disease, peripheral artery disease (PAD), and other forms of ASO as unique snowflakes. There are, of course, circumstances where it makes sense to do this: angina, stroke, and claudication are obviously different problems with different complications, prognoses, and treatments. Coronary artery stents relieve angina. Carotid endarterectomy prevents stroke. A walking program helps claudication. It therefore makes perfect sense to split these diseases into their own categories and emphasize their differences. Right?

Not so fast. I have another friend who objects to the current approach of naming different forms of ASO according to the arteries and organs affected. He would prefer to emphasize that a single disease – atherosclerosis – is at the root of the problem regardless of the organs affected or symptoms produced. As he explains it:

Just contrast the way we describe atherosclerosis as opposed to cancer. Atherosclerosis is a single, unique disease, yet it is usually broken into numerous disease types based on the tendency of a particular blocked artery to cause a distinct symptom. In contrast, each type of cancer really is unique to the organ or tissue from which it arises – it is literally the uncontrolled expansion of a distinct cell line – but the many manifestations are commonly thought of as one disease, and they all contain ‘cancer’ in the name (breast cancer, colon cancer, lung cancer). For example, lung cancer that metastasizes to the brain isn’t ‘brain cancer’. It’s still lung cancer, but the symptoms are different from lung cancer that metastasizes to the spine. The same is true for atherosclerosis. It’s the same disease regardless of whether the symptoms involve the heart, brain, legs, or something else. (J. Beckman, personal communication, 2017.)

My lumper friend remembers things about ASO that splitters occasionally forget: if you have one form of atherosclerosis, you have them all (so even if your only symptom is claudication, you’re still likely to die from occult coronary or carotid artery disease). Regardless of the specific form of ASO that predominates, you need to stop smoking, control your hypertension, take a statin, and watch your blood sugar. Exercise is good. Obesity is bad. And so on. As vascular specialists we know that these types of common elements are important. Yet some of our colleagues in sister disciplines have become so focused on individual subtypes of ASO that they forget everything they’ve ever learned about other forms of the disease – experts in coronary or carotid artery disease can seem surprisingly clueless when it comes to, for example, PAD.

Varicose versus collateral veins – splitting based on presumed etiology

Everybody knows that varicose veins are the result of unhealthy venous degeneration caused by genetic flaws, chronic venous hypertension, inflammation, traumatic injury, and other factors. They bulge, they’re tender, and they look ugly. Varicose veins are inherently bad and you certainly don’t want them, right? In contrast, the collateral veins that develop around chronic venous occlusions are the product of healthy vascular growth and proliferation. Sure, they can also be ugly and tender – just like varicose veins – but if your venous outflow is obstructed, you certainly want to grow these. It seems obvious that varicose veins and venous collaterals are different things … except that maybe they’re not. ⁷ Varicose veins and venous collateral veins turn out to be anatomically and histologically similar. They develop in response to the same hormones and growth factors. Both appear to form in response to the same intrinsic angiogenic capabilities. It’s assumed that varicose veins and collateral veins are entirely different and reflect different etiologies, but perhaps they’re actually similar structures with a common etiology. Maybe we should lump rather than split them for study purposes?

As vascular medicine specialists we are, by nature, splitters. That’s not such a bad thing. Splitting offers a valuable approach for exploring and understanding subtle differences between various forms of vascular diseases. Splitting allows us to carve out discrete areas of expertise that are not just unique, but clinically useful. Splitting enables us to differentiate ourselves from general internists, cardiologists, vascular surgeons, and other groups that lay claim to the intellectual and professional territory where blood vessels reside. I will publicly, if somewhat reluctantly, admit that splitting is a parade upon which I should not rain. But I’ll also remind folks that we need to avoid focusing so narrowly on the differences between various vascular diseases that we lose sight of the similarities they share. Why? Because vascular diseases are like the people they afflict. The differences between them are important, but the similarities are essential.