In this series of 201 autopsy cases of TDP‐43 proteinopathies, many patients had concurrent tau pathologies. Not unexpectedly, ARTAG (42%) and PART (36%) were the most frequent tau pathologies, followed by AGD (22%). In addition, we found two patients with CBD, which is a four‐repeat tauopathy form of FTLD‐tau. TDP‐43 and tau are the most common molecular subtypes of FTLD, but the coexistence of FTLD‐TDP and FTLD‐tau is uncommon [23]. These cases raise the issue of which should be considered the ‘primary’ pathologic process and which should be considered the ‘secondary’ process.

Case 1 was clinically diagnosed with familial AD based upon his cognitive impairment, initially characterised by amnestic type dementia, as well as dementia in multiple family members. Neuropathologic assessment revealed mild Alzheimer's‐type pathology (Braak NFT stage III and Thal amyloid phase 3) insufficient to account for dementia. Although the cortical atrophy of the frontal and temporal lobes was relatively mild, immunohistochemistry for TDP‐43 and presence of C9RANT inclusions were diagnostic of FTLD‐TDP [38]. A hexanucleotide repeat expansion in the C9orf72 gene was confirmed with repeat‐primed polymerase chain reaction assay [35]. The presence of a pathogenic mutation in a gene for FTLD makes a strong case for the primary diagnosis in the case to be FTLD‐TDP. Of note, amnestic Alzheimer's dementia is a common clinical diagnosis of genetically confirmed FTLD‐TDP in elderly individuals [39, 40], and in an autopsy series from the State of Florida brain bank, late onset patients with C9orf72 mutations often present with Alzheimer type dementia or Lewy body dementia [41].

Tau pathology in this patient was consistent with CBD based upon morphology and neuroanatomical distribution; however, tau pathology was mild and subcortical nuclei vulnerable to neuronal loss in CBD, such as the globus pallidus and substantia nigra, were well preserved. Moreover, the severity of tau pathology was mild in subcortical regions. These findings are like those reported as ‘preclinical’ CBD [42]. In the current situation, the term ‘preclinical’ is not appropriate since the patient presented with cognitive impairment, behavioural changes and parkinsonism. Taken together, genetically confirmed FTLD‐TDP with ‘incidental’ CBD seems to be the best neuropathologic diagnosis.

Interestingly, this patient is like a patient in the study of Kim and co‐workers. Their patient had FTLD‐TDP type A and unclassifiable FTLD‐tau with C9orf72 mutation (case 1) [23]. Given that C9orf72 mutation is the most common risk factor for familial amyotrophic lateral sclerosis and FTLD, its strong association with TDP‐43 pathology has been established. In contrast, it is still unknown whether C9orf72 mutation is also associated with FTLD‐tau. Bieniek et al [43]. investigated Alzheimer's‐type tau pathology in the temporal cortex and hippocampus in patients with FTLD who carried C9orf72 mutation (c9FTLD) and found that tau pathology burden was not different between c9FTLD and sporadic FTLD. Snowden et al [44]. screened for C9orf72 mutations in 398 patients with clinical presentations of frontotemporal dementia and found one patient who had CBD pathology without TDP‐43 pathology. The absence of TDP‐43 pathology raised the question of the significance of C9orf72 hexanucleotide expansion in this patient. King et al [45]. reported a patient with C9orf72 mutation who had Pick's disease and TDP‐43 pathology, with TDP‐43 co‐localising with Pick bodies. This patient also had a MAPT variant A239T, which has not been associated with tau pathology. Taken together, the association between C9orf72 mutation and tau pathology remains uncertain and warrants further investigation.

Our second case had marked atrophy in frontal and temporal lobes, as well as severe atrophy in the orbital gyrus and caudate nucleus, and hippocampal sclerosis consistent with FTLD‐TDP. In contrast, marked pigment loss in the substantia nigra and discolouration of the globus pallidus were suggestive of CBD. TDP‐43 immunohistochemistry revealed moderate NCI and dystrophic neurites and sparse neuronal intranuclear inclusions in the neocortex, consistent with FTLD‐TDP type A. There was also severe tau pathology in the grey and white matter of the neocortex, deep grey nuclei, brainstem and cerebellum, consistent with CBD. Given that both FTLD‐TDP and CBD can present as behavioural variant frontotemporal dementia, it is difficult to determine which pathology is ‘primary’ for this patient. One can argue that CBD is the ‘primary’ diagnosis in this patient and that TDP‐43 pathology can be explained by limbic‐predominant age‐related TDP‐43 encephalopathy neuropathological change (LATE‐NC) [46]. LATE‐NC is characterised by TDP‐43 proteinopathy in the medial temporal lobe in the elderly with or without hippocampal sclerosis [46]. The amygdala and hippocampus are the most vulnerable regions, but the midfrontal gyrus is affected in stage 3; therefore, it is challenging to differentiate FTLD‐TDP and advanced stages of LATE‐NC [47]. Case 2 had significant frontal lobe atrophy, which supports the diagnosis of FTLD‐TDP; however, frontal lobe atrophy is also common in CBD. The advanced age of the patient (84 years) and the presence of hippocampal sclerosis may also support the diagnosis of LATE‐NC. In this sense, a diagnosis of CBD with LATE‐NC could be reasonable. Nevertheless, our final neuropathological diagnosis is FTLD‐TDP with ‘incidental’ CBD because atrophy of the globus pallidus and subthalamic nucleus were relatively mild compared to that of the orbital gyrus and caudate nucleus. We judge the CBD to be ‘incidental’ pathology in the context of FTLD‐TDP.

The present study needs to consider the fact that TDP‐43 pathology can be found in a subset of CBD cases. Our previous study identified astrocytic plaque‐like TDP‐43 lesions in the motor cortex and superior frontal gyrus in CBD patients with TDP‐43 pathology. Therefore, we assumed that TDP‐43 pathology was ‘secondary’ to CBD pathology [19]. In contrast, the two patients in the present study did not have astrocytic plaques with TDP‐43 immunoreactive processes. This finding supports the conclusion that TDP‐43 pathology occurs independently from tau pathology in CBD, not ‘secondary’ to tau pathology.

A limitation of this study is that the clinical information of two patients with FTLD‐TDP and CBD was limited due to the retrospective nature of the autopsy cohort. Systematic, longitudinal data with neurological examinations and neuropsychological testing were not available. Motor symptoms suggestive of corticobasal syndrome may have been overlooked or not well documented. Another limitation is that we did not assess clinical features in patients with PART, ARTAG or AGD because the primary focus was FTLD‐tau, rather than these ‘age‐related’ tau pathologies.

In conclusion, by screening for tau pathology in TDP‐43 proteinopathies, we identified two patients with mixed FTLD‐TDP and CBD that are likely independent (‘co‐primary’) disease processes. Many of elderly individuals with neurodegenerative disorders have multiple coexisting pathologies [11, 12]; therefore, both TDP‐43 and tau should be part of the screening process for the neuropathologic assessment of FTLD. We are not able to determine the relative contribution of each pathology to the clinical presentations in these patients. This is not too dissimilar to the issue of assigning relative contributions of mixed pathology in other settings, for example, cases with both Alzheimer's disease and diffuse Lewy body disease. Implementation of molecular imaging modalities for each protein may help determine the timing and relative contributions of the proteins to clinical presentations.