Primary progressive aphasia (PPA) is diagnosed when a progressive language impairment (fluent or nonfluent) emerges in relative isolation during the initial stages of a degenerative disease that eventually leads to an aphasic dementia.1, 2 Knibb and colleagues'3 recent article offers another important contribution to this area by the Cambridge group. Using my PPA criteria,2, 4 but liberalizing the 2-year period of symptom specificity, they showed that approximately two of three cases have the neuropathologies of frontotemporal lobar degeneration, and that the aphasia subtype influences the frequency of ubiquitin versus tau inclusions.

A third of the patients had Alzheimer's disease (AD) pathology. This led to my being quoted in Knibb and colleagues'3 article as follows: “Mesulam has said that ‘the neuropathological diagnosis of AD in a patient with the clinical picture of PPA…should be met with skepticism; such a diagnosis would have failed to establish a credible clinicopathological correlation.’” The missing phrase, replaced by an ellipsis, had advised skepticism only if there was AD pathology “and a distribution of neurofibrillary tangles and amyloid plaques typical of AD.”4 Because neurodegeneration is initially maximal in the hippocampal/entorhinal area of typical AD, my statement had cautioned that such pathology was incompatible with the “aphasia without amnesia” pattern of PPA. Is the AD pathology of PPA “atypical”? If so, through what mechanism?

According to the title, “Clinical and Pathological Characterization of Progressive Aphasia,” Knibb and colleagues3 consider semantic dementia (SD) and progressive nonfluent aphasia (PNFA) to be aphasic disorders. These syndromes can therefore be considered PPA subtypes, just as Wernicke's and Broca's are subtypes of stroke-induced aphasia. The differential inclusion frequency (43% tau in PNFA vs 13% in SD; 53% ubiquitin in SD vs 17% in PNFA; nearly 30% AD in each) does not alter this conclusion. What is remarkable is the overlap. In support of unification, the codon 129 MV polymorphism in the PrP gene is overrepresented in PPA (but not AD) regardless of aphasia subtype.5

PPA can be resolved into three subtypes: agrammatical (abnormal syntax and fluency; ie, PNFA), semantic (poor word comprehension; ie, SD without prominent visual agnosia), and logopenic (word-finding pauses without agrammatism or word comprehension deficits). Gorno-Tempini and coworkers,6 who introduced this classification, described a subtype-specific anatomy. Knibb and colleagues3 suggest that these patterns may also reflect differential neuropathological associations.

Neary and coworkers7 list face/object recognition impairments as core features of SD, whereas Knibb and colleagues3 do not appear to, making it easier to bring SD into the fold of aphasias. Future articles should specify which definition of SD is being used. In the meantime, Knibb and colleagues3 promote a coherent conceptualization of this fascinating syndrome where the language network becomes the preferential target of neurodegeneration.


  1. Top of page

Marsel Mesulam MD*, * Cognitive Neurology and Alzheimer's Disease Center, Northwestern University Feinberg School of Medicine, Chicago, IL