The primary finding of the present study is the lack of a significant impact of the tracer, 99mTc-HMPAO or 99mTc-ECD, on the performance of ictal brain perfusion SPECT for identification of the SOZ in patients with drug resistant epilepsy. Neither the proportion of lateralising ictal SPECT nor the 12 months outcome after resection in the lobe pointed to by ictal SPECT differed significantly between the 99mTc-HMPAO and the 99mTc-ECD group. These findings do not rule out small to medium sized (clinically relevant) differences, despite the fact that the sample size was considerably larger in the present study compared to the previous studies that directly compared 99mTc-HMPAO and 99mTc-ECD for ictal brain perfusion SPECT [25, 26]. The power of the present study to detect a difference of 5, 10, 15, 20 and 25% in the proportion of lateralising SPECT between 99mTc-HMPAO and 99mTc-ECD (65% versus 70, 75, 80, 85, and 90%) at the two-sided 5% level was 9%, 28%, 59%, 87% and 99%, respectively.
The lack of a significant tracer effect in the present study at first sight appears to be in conflict with the two previous studies that directly compared 99mTc-HMPAO and 99mTc-ECD for ictal perfusion SPECT and reported findings in favor of 99mTc-HMPAO  or in favor of 99mTc-ECD . However, the sensitivity estimates reported in the study favoring 99mTc-HMPAO counted cases with temporal hyperperfusion in the wrong hemisphere (according to invasive EEG or surgical outcome as standard of truth) as true positive . This approach resulted in significantly higher sensitivity with 99mTc-HMPAO compared to 99mTc-ECD (70% versus 29%, p = 0.03) in the 30 patients with extratemporal epilepsy (23 with 99mTc-HMPAO, 7 with 99mTc-ECD). When only cases with correct lateralization were counted as true positive, the sensitivity of 99mTc-HMPAO dropped to 35% and most likely did not differ significantly from the sensitivity of 99mTc-ECD (not reported in the publication). The study that reported results in favor of 99mTc-ECD included a significantly higher proportion of postictal scans (tracer injection after seizure activity had ended) for 99mTc-HMPAO than for 99mTc-ECD (57% versus 16%, p < 0.0001) . This was caused by an additional average delay of 40 s for reconstitution of unstabilized 99mTc-HMPAO after seizure onset . When the analysis was restricted to patients with ictal injections, there was no significant difference in the localization rate between 99mTc-HMPAO and 99mTc-ECD (85% versus 89%) . Thus, closer inspection of the two previous comparisons of 99mTc-HMPAO and 99mTc-ECD for ictal perfusion SPECT reveals that they do not provide strong evidence for the superiority of 99mTc-ECD over 99mTc-HMPAO or vice versa. A meta-analysis of SPECT brain imaging in epilepsy reported that the performance of SPECT was not dependent on the tracer, but the meta-analysis did not include 99mTc-ECD SPECT studies but only 5 series with a total of 69 patients in whom an 123I-labelled tracer (123I-HIPDM or 123I-IMP) had been used for ictal perfusion SPECT instead of 99mTc-HMPAO .
A secondary finding of the present study was the 63% sensitivity of ictal perfusion SPECT (independent of the tracer) for correct localization (including correct lateralization) in 62 patients with “good” or “improved” outcome 12 months after temporal epilepsy surgery. This is lower than reported by most previous studies. For example, a prospective multi-center study initiated by the International Atomic Energy Agency in patients with temporal lobe epilepsy undergoing surgery found sensitivity of 86.5% and 83.8% for correct lateralization and localization (in the temporal lobe) of ictal 99mTc-ECD SPECT taking the surgical site as the gold standard . The meta-analysis cited above reported the combined sensitivity of unifocal findings in ictal perfusion SPECT ipsilateral to the operative side among all good surgical outcomes to be 96.7% (95%-confidence interval 88.7–99.6%) . Possible explanations of the lower sensitivity of ictal SPECT in the present study include differences in the patient population, particularly a lower proportion of patients with seizures that were already well localized by other presurgical methods (as expected in a sample from clinical routine patient care without specific eligibility criteria). In line with this, the proportion of patients with non-lateralizing MRI was rather high in the present study (Table 1).
Another possible explanation for the comparatively low sensitivity of ictal SPECT in the present study is the rather high cutoff of 120 s on the latency of tracer injection applied for exclusion of patients, although the latency was longer than 60 s in only 4 of the 172 patients with known latency. The rationale for the 120 s cutoff was that in temporal lobe epilepsy hyperperfusion in mesial temporal structures is observed up to 2 minutes after the ictus, whereas hypoperfusion of the whole temporal lobe (“postictal switch”) is observed from 2–15 min after the ictus [27, 28]. However, after intravenous injection, the tracer takes about 40 s to reach the brain and to be metabolized to become locally fixed . As a consequence, even with very early injection (latency ≤ 30 s) the SPECT image might primarily represent postictal perfusion in case of short seizures (≤ 30 s) . In order to address this point, an additional analysis was performed in the subgroup of patients with “good” or “improved” outcome at 12 months after temporal epilepsy surgery with latency of tracer injection ≤ 60 s and electrical duration of the seizure after tracer injection ≥ 20 s (n = 50). Lateralization of the ictal perfusion SPECT was correct in 34 of the 50 patients, corresponding to 68% sensitivity, that is, slightly better than the 63% sensitivity of correct lateralization in all patients with “good” or “improved” outcome at 12 months after temporal epilepsy surgery.
The following limitations of the study should be mentioned. First, the study was retrospective, which explains the rather high fraction of missing data for some variables (Table 1). In order to avoid potential selection bias by the retrospective inclusion, only very liberal eligibility criteria were applied. As a consequence, the included patient sample should be representative of patients referred to ictal perfusion SPECT for presurgical evaluation of suspected focal epilepsy in clinical routine. Second, 99mTc-HMPAO and 99mTc-ECD had not been used concurrently, but 99mTc-HMPAO SPECT had been performed after 99mTc-ECD SPECT with only a few exceptions. In order to avoid a potential time effect (e.g., by varying experience of the personnel), processing and visual interpretation of all SPECT images was performed retrospectively by the same readers according to a standardized protocol (instead of using the original interpretation in the written report in the patients’ files). All SPECT had been acquired with the same SPECT systems and using the same acquisition and reconstruction protocol independent of the tracer. Third, the 99mTc-HMPAO group and the 99mTc-ECD group were well matched with respect to age, sex, age at first seizure, duration of disease, seizure frequency, history of previous brain surgery, and findings in presurgical MRI (Table 1). The tracer groups differed significantly with respect to the latency of tracer injection (median latency 4 s longer in the 99mTc-HMPAO group), the duration of the seizure after tracer injection (25 s shorter in the 99mTc-HMPAO group), the tracer dose (70 MBq higher in the 99mTc-HMPAO group), and with respect to the delay of the SPECT acquisition after tracer injection (63 min longer in the 99mTc-HMPAO group). The difference of the median latency of tracer injection between the groups of 4 s was rather small and probably would not have reached statistical significance with smaller sample size. However, it cannot be ruled out that it had a relevant impact on the results, although this appears unlikely. The difference in tracer dose also is not expected to have significantly affected the results . The same is true for the difference in the delay of the SPECT acquisition after tracer injection, as repeat SPECT imaging at 30 min, 2 h and 4–7 h after a single injection of 99mTc-ECD in patients with temporal lobe epilepsy did not reveal a relevant impact of the delay between tracer injection and start of the SPECT acquisition on the perfusion pattern . The rather large group difference with respect to the delay of the SPECT acquisition after tracer injection is explained by the fact that the transport from ictal tracer injection in the inpatient epilepsy unit of the Department of Neurology and Epileptology of the Protestant Hospital Alsterdorf to the SPECT acquisition in the Department of Nuclear Medicine of the University Medical Center Hamburg Eppendorf was by taxi in the majority of 99mTc-ECD cases versus ambulance patient transport in the majority of 99mTc-HMPAO cases. Finally, the two tracer groups comprised different patients instead of head-to-head comparison of the two tracers in the same patients. Head-to-head comparison of the two tracers in the same patients is limited by the fact that both tracers are radioactively labeled by the same isotope (99mTc) so that they cannot be used simultaneously during the same seizure. Sequential SPECT acquisition during different seizures of the same patient is affected by considerable intrasubject variability of the perfusion pattern in ictal SPECT between different seizures reflecting methodological differences (e.g., different injection latency) but also differences between seizures of the same patient [32, 33].
In conclusion, this study does not provide evidence to favor 99mTc-HMPAO or 99mTc-ECD for identification of the seizure onset zone by ictal perfusion SPECT in patients with drug resistant epilepsy.