Is the SPECT FTD pattern indicative of developing FTD over time in populations with suspected dementia? What is the diagnostic accuracy of rCBF SPECT biomarker for discriminating FTD from non‐FTD, and FTD from Alzheimer's disease dementia and other dementia subtypes?

Population

Participants with suspected dementia and rCBF SPECT administered at baseline (prospective cohort design) (n = 3)

Participants with suspected dementia and rCBF SPECT administered at baseline with histopathological confirmation (retrospective cohort design) (n = 2)

Participants clinically diagnosed with FTD or other dementia subtypes using standard clinical diagnostic criteria (case‐control design) (n = 6)

Setting

Outpatients from University centres or University memory clinics (n = 7)

Outpatients from General Hospital memory clinics (n = 1)

Tertiary referral centre (n = 1)

Multicentre (different French hospitals) (n = 1)

Not reported (n = 1)

Sampling procedure

Consecutive or random (n = 3)

Not consecutive or random (n = 4)

Not reported (n = 4)

Prior testing

Prior to performing rCBF SPECT scans, diagnostic criteria for identifying dementia subtypes were applied in studies that used a case‐control study design

Index tests

^99mTc‐HMPAO SPECT; ^99mTc‐ECD SPECT; Xenon SPECT

Threshold prespecified at baseline

Yes (n = 6)

No (n = 3)

Unclear (n = 2)

Threshold

Included studies used a range of thresholds

SPECT scan interpretation

Combined visual and semiquantitative interpretation (n = 6)

Visual interpretation only (n = 5)

rCBF hypoperfusion region

Authors used brain regions that were expected to be affected by FTD and so frontal and/or temporal lobes were involved in all studies. Two studies also included parietal and occipital lobes in their evaluations. One study used a range of Broadmann areas (BAs)

Target condition

Frontotemporal dementia (FTD): ante‐mortem clinical diagnosis of FTD (n = 7) or neuropathological diagnosis of FTD (n = 4)

Reference standard

For ante‐mortem clinical diagnosis: Neary 1998 criteria (n = 4); Brun 1994 criteria (n = 1); not specified (n = 2)

For neuropathological diagnosis: Shi 2005 (n = 1); Cairns 2007 (n = 1); not specified (n = 2)

Diagnostic criteria for dementia subtypes

For AD dementia: NINCDS‐ADRDA criteria (n = 6); not specified (n = 1); histopathological criteria (n = 4)

For vascular dementia: NINDS‐AIREN criteria (n = 2); histopathological (n = 1)

Included studies

Eleven studies (n = 1077 participants) assessed rCBF SPECT for differentiating between FTD and AD. Five of these studies (n = 609) also assessed rCBF SPECT for differentiating between FTD and non‐FTD

Reference standard: neuropathological diagnosis. Objective A: rCBF SPECT FTD type pattern (at baseline) indicative of FTD (at follow up) in participants with suspected FTD at baseline; Objective B: The accuracy of rCBF SPECT pattern in differentiating FTD from AD

Objective

Study

N

Confirmed

FTD

Sensitivity

(95% CI)

Specificity

(95% CI)

Quality

Comment

A

Read 1995

27

7

100%

(0.59, 1.00)

100%

(0.83, 1.00)

Unclear risk of bias was seen in the patient selection QUADAS‐2 domain. The remaining three domains considered to be at low risk of bias. There were no concerns about applicability (Read 1995; Rollin‐Sillaire 2012).

High risk of bias was seen in patient selection, index test, and flow and timing QUADAS‐2 domains. The reference standard was strength of the study. There were no concerns about applicability (Lipton 2004)

High risk of bias was seen in participant selection and index test domain. The reference standard was strength of the study. There were no concerns about applicability (McNeill 2007)

Objective A: These papers used the gold standard of histopathological diagnosis; however. the methods used in participant selection and image analysis have led to the introduction of a degree of bias

Objective B: These papers used the gold standard of histopathological diagnosis. Although the diagnosis is robust, case‐control design, small study numbers, different methodologies with a wide range of sensitivities and specificities mean that it is difficult to make recommendations on the basis of these results

A

Rollin‐Sillaire 2012

48

9

75%

(0.43, 0.95)

97%

(0.85. 1.00)

B

Read 1995

20

7

100%

(0.59, 1.00)

100%

(0.75, 1.00)

B

Rollin‐Sillaire 2012

35

9

75%

(0.43, 0.95)

100%

(0.85, 1.00)

B

Lipton 2004

23

6

100%

(0.54, 1.00)

41%

(0.18, 0.67)

B

McNeill 2007

56

20

80% (0.59, 0.93)

61 % (0.42, 0.78)

Investigation

of heterogeneity

We were not able to formally assess the effect of potential sources of heterogeneity because meta‐analyses were not performed

Conclusion

Further research on the use of rCBF SPECT for differentiating FTD from other dementias is required. In particular, protocols should be standardised, study populations well described, threshold for 'abnormal' scans predefined and clear details given on how scans are analysed