A harmonized classification system for FTLD-TDP pathology

In 2006, two papers were published, each describing pathological heterogeneity in cases of frontotemporal lobar degeneration (FTLD) with ubiquitin-positive, tau-negative inclusions (FTLD-U) [7, 11]. In both studies, large series of cases were evaluated and the investigators felt that they could recognize three distinct histological patterns, based on the morphology and anatomical distribution of ubiquitin immunoreactive neuronal inclusions. The findings of Sampathu et al. were further supported by differential labelling of the pathology, using a panel of novel monoclonal antibodies; whereas, Mackenzie et al. found relatively specific clinicopathological correlations. Most importantly, the pathological features that defined the subtypes in these two studies were almost identical, providing powerful validation of the results. However, because the studies were conducted simultaneously and independently, the numbering of the subtypes, used in the respective papers, did not match (Table 1).

Table 1.

Proposed new classification system for FTLD-TDP pathology, compared with existing systems

New system	Mackenzie et al. [7]	Sampathu et al. [11]	Cortical pathology	Common phenotype	Associated genetic defects
Type A	Type 1	Type 3	Many NCI Many short DN Predominantly layer 2	bvFTD PNFA	GRN mutations
Type B	Type 3	Type 2	Moderate NCI Few DN All layers	bvFTD MND with FTD	Linkage to chromosome 9p
Type C	Type 2	Type 1	Many long DN Few NCI Predominantly layer 2	SD bvFTD
Type D	Type 4a	Type 4a	Many short DN Many lentiform NII Few NCI All layers	Familial IBMPFD	VCP mutations

bvFTD behavioural variant frontotemporal dementia, DN dystrophic neurites, GRN progranulin gene, IBMPFD inclusion body myopathy with Paget’s disease of bone and frontotemporal dementia, MND motor neuron disease, NCI neuronal cytoplasmic inclusions, NII neuronal intra-nuclear inclusions, PNFA progressive non-fluent aphasia, SD semantic dementia, VCP valosin-containing protein gene

^aDescribed subsequently by Forman et al. [4]

Shortly thereafter, further work by one of the two groups led to the identification of the transactive response DNA-binding protein with M_r 43 kD (TDP-43) as the ubiquitinated pathological protein in most cases of FTLD-U as well as the majority of sporadic amyotrophic lateral sclerosis (ALS) and some familial ALS [10]. It was subsequently confirmed that most FTLD-U cases had TDP-43 pathology and that the same pathological patterns could be recognized based on the results of TDP-43 immunohistochemistry (IHC) [1, 2]. By this time, a fourth FTLD-U subtype had been described, specifically associated with the familial syndrome of inclusion body myopathy with Paget’s disease of bone and frontotemporal dementia (IBMPFD) caused by mutations in the valosin-containing protein (VCP) gene [4], and this was also shown to have TDP-43 pathology [9]. As a result, cases of FTLD with TDP-43 pathology are now designated as FTLD-TDP and the term FTLD-U is no longer recommended [8].

The two classification systems for FTLD-U/FTLD-TDP have now gained wide acceptance and have repeatedly been validated by the discovery of additional clinical, genetic and pathological correlations. However, the continued use of two discordant numbering systems proves to be an ongoing source of confusion within the field. Previous attempts, by other groups of authors, to promote one classification over the other have not been successful.

To resolve this issue, the principal authors of the original two papers are now proposing a new classification for FTLD-TDP pathology, the sole purpose of which is to provide a single harmonized system that replaces the two currently in use. In developing this new classification, the following principles were adhered to: (1) different pathological subtypes are designated by letters to help distinguish this from the pre-existing number-based systems, (2) the order of subtypes should not exactly match either of the previous systems to avoid any apparent bias, and (3) the order of the subtypes should be based on their relative frequency, with “A” being the most common.

The result is summarized in Table 1. Type A is equivalent to type 1 of Mackenzie et al. and type 3 of Sampathu et al., being characterized by numerous short dystrophic neurites (DN) and crescentic or oval neuronal cytoplasmic inclusions (NCI), concentrated primarily in neocortical layer 2. Moderate numbers of lentiform neuronal intranuclear inclusions (NII) are also a common but inconsistent feature of this subtype. Type B matches Mackenzie et al. type 3 and Sampathu et al. type 2, with moderate numbers of NCI, throughout all cortical layers, but very few DN. Type C is the same as Mackenzie et al. type 2 and Sampathu et al. type 1, having a predominance of elongated DN in upper cortical layers, with very few NCI. Finally, Type D refers to the pathology associated with IBMPFD caused by VCP mutations, characterized by numerous short DN and frequent lentiform NII.

Based on the results of more recent studies, there are a number of other modifications that we could have considered incorporating into this new system. Additional pathological subtypes could be added; for instance, to describe the TDP-43 pathology that is found in the mesial temporal lobe in a high proportion of cases of Alzheimer’s disease and most other common neurodegenerative conditions [3]. The pathological criteria for each of the subtypes could be expanded to include characteristic findings in subcortical regions [5, 6]. The description of the pathological features could be modified to take into account the greater sensitivity and specificity of TDP-43 IHC, which may demonstrate additional findings, not recognized with the ubiquitin immunostaining techniques upon which the original classifications were based (such as neuronal “pre-inclusions”) [2]. Although these and other recent findings represent important advances in our understanding of FTLD-TDP, most have not yet been broadly replicated or completely defined. Therefore, in order to make the transition to a new classification as simple and widely acceptable as possible and, most importantly, to allow for direct translation with the currently existing systems, we are not proposing any other significant changes, beyond the coding of the subtypes.

In summary, we believed that adoption of a single harmonized system for the classification of FTLD-TDP neuropathology would greatly improve communication within the rapidly advancing field of FTLD diagnosis and research. Future attempts to resolve any outstanding issues related to the practical implementation and interpretation of FTLD pathological classification should also benefit. As indicated by their inclusion as co-authors on this paper, this proposal has received the unanimous support of all of the neuropathologists involved in the original two studies [7, 11].