The clinical acceptability of short versus long duration acquisitions for head and neck cancer using long-axial field-of-view PET/CT: a retrospective evaluation

The principle that glucose metabolism, as measured by the uptake of the positron emission radiotracer fluorodeoxyglucose (2-[¹⁸F]-FDG), could be used to measure the proliferative activity of head and neck cancers (HNC) was first established by Minn et al. in 1988 [1] who subsequently established that changes in uptake correlated with treatment outcome after radiotherapy [2]. Accordingly, combined positron emission and computed tomography (PET/CT) has become a well-established tool in the staging and evaluation of HNC, particularly in advanced stage disease [3]. Now in its third decade, PET/CT continues to undergo rapid and important technological development. For example, the introduction of time-of-flight and fully-digital systems have furnished improvements in scanner sensitivity and resolution, with faster time of flight resolution, increased scanner sensitivity and image quality [4]. These systems can improve diagnostic performance [5‐7].

Most recently, long-axial field of view (LAFOV) systems have been introduced onto the market. They exhibit substantially improved whole-body coverage, sensitivity and quantification, and can afford low-activity or ultra-fast scan protocols [8‐13]. The total recovered signal from a lesion is a function of both applied activity and acquisition time as well as dependent upon scanner characteristics [14]. However, the clinical utility of any substantially higher sensitivity on lesion detectability has not yet been fully evaluated. For example, it has been hypothesised that the higher scanner sensitivity might yield higher detection rates or afford detection of pathology at earlier stages, as has previously been demonstrated for digital scanners [5], e.g. in the detection of micrometastases in lymph nodes (LN) [15, 16]. Alternatively, the higher sensitivity can be exploited to reduce scan time. For example, Sachpekidis et al. (2022) demonstrate in patients with malignant melanoma that a 5-min low-dose acquisition using a LAFOV system can deliver comparable results to a standard scan [17]. The optimal acquisition time and activity for HNC in LAFOV is yet to be established.

A number of early studies using analogue PET systems demonstrated the additional clinical value of a high-resolution acquisition of the head and neck (H&N) region, with improved LN Detection [18‐20] and this is common practice in many centres [21], and our institutional standard when using our digital SAFOV system (Siemens Biograph Vision) is for a 4-min per bed position equivalent acquisition over the head and neck in continuous bed motion.

The ability to scan the entire torso in a single bed position using LAFOV coupled with improved detection geometries allows for acquisitions which would be unfeasibly long when using a short axial FOV (SAFOV) system. To yield equivalent results for our clinical standard 10-min acquisition on the LAFOV, an impracticable 88 min using the SAFOV Siemens Biograph Vision in continuous bed motion (CBM) would be required for equivalent FOV coverage. This coupled with the higher sensitivity of a LAFOV system owing to improved geometry results in substantially higher count statistics with low noise images. In determining the optimal acquisition protocol for HNC in LAFOV, the balance between a clinical benefit in scanning longer or with more activity with higher image quality must be balanced against the requirement to scan faster or more gently, with lower activity protocols.

The aim of this study was therefore to assess the clinical acceptability of standard-duration (4 min, C_40%) acquisitions compared to full-count (10 min, C_100%) long duration acquisitions using a LAFOV system in terms of the detection of LN metastases in HNC. In doing so, we also assess the potential additional clinical value of the long-duration acquisitions which LAFOV is now able to provide and to evaluate any differences in diagnostic accuracy therein.

In this study, we find that a 4-min (C_40%) acquisition and a 10-min (C_100%) acquisition using standard activities on a LAFOV system show no significant differences in terms of lymph node detection in patients undergoing PET/CT for HNC. This demonstrates that, compared to standard activities and scan duration, substantial reductions in both are feasible and clinically acceptable with LAFOV systems. Whereas half-dose acquisitions with LAFOV have recently been shown to be acceptable in lung cancer [26], and low-dose 5-min acquisitions are suitable for melanoma [17], the optimal scan duration or applied activity for HNC with LAFOV is yet to be determined. This present study represents the first systematic evaluation of LAFOV PET/CT in HNC and is the first to assess the clinical value of the low noise high integral count activity examinations.

Alternatively, a 10-min scan could be performed and only 40% of the activity be applied, resulting in a reduction in radiation exposure from 4.3 to 1.7 mSv for a standard 75 kg adult patient [22], and further activity reductions might be possible without detriment to image quality [17]. Faster scans also might result in reduced patient movement, be more acceptable to patients with claustrophobia and allow for greater scanner throughput, widening access and shortening waiting times [27].

Although LAFOV systems allow for faster or gentler imaging with lower activities owing to their substantial higher sensitivity profiles through improved geometry [11, 28], the degree to which this additional scanner sensitivity allows for better or more accurate imaging is yet to be established. As was discussed earlier, longer-duration dedicated H&N-acquisitions on analogue systems showed improved detection of LN metastases in HNC [18‐20]. Congruent to this finding, it might be postulated that the higher sensitivity of LAFOV systems might result in yet further improvements in diagnostic performance. For example, LAFOV might allow micro-metastases to be detected [15, 16]. Earlier and more accurate detection of pathology has already been demonstrated for digital silicon photomultiplier systems when compared to analogue systems [5‐7, 29] but these findings are yet to be replicated with LAFOV systems. Although not powered as a superiority study, we did indeed find small but non-significant improvement in LN detection in the full count (C_100%) scans, with additional lesions identified in 32 patients. The clinical vignettes shown in Fig. 2 show how the higher sensitivity afforded by full-count scans allow better detection of small LN metastases in HNC. Likewise, the lower noise at C_100% afforded better interpretation of a LN with low but focal avidity at the C_40%. Follow-up imaging post radiotherapy, where the LN was outside of the radiation port and remained unchanged confirmed that this finding was a reactive lymph node. Likewise, small but borderline-significant improvements were observed in C_100% acquisitions for accuracy and ROC-analysis revealed a lower SUV cut-off. Taken together, we interpret these findings as suggestive that longer, full-count acquisitions provided some additional diagnostic benefit which translate to improved clinical performance. In-keeping with previous studies in digital PET, we find that this higher detection does translate to lower reader confidence with higher rates of equivocal findings at C_100% [5], although it is important to note that this was not at the expense of accuracy. When devising fast or low-activity protocols with resultant lower lesion integral activities, care should be taken that the potential benefits of high-quality full-count and low noise acquisitions are not forfeited.

The dual-time point evaluation of uptake kinetics has been investigated and might help differentiate between LN metastases and inflammatory LN in a number of entities [30‐34]. Twenty-minute duration acquisitions have already been demonstrated to allow reliable analysis of tracer kinetics [35, 36]. We therefore wondered whether the high temporal resolution which LAFOV provides could provide insight into tracer kinetics over 10 min [35]. However, our data show no appreciable changes in SUV_max. Conversely, reliable quantification was possible even in the space of a 4 min scan, where SUV_max can vary with noise and exposure [14].

This study has some limitations. The present study compares acquisitions after 4 min (chosen as the equivalent to a clinical standard SAFOV scan) and the full 10-min acquisition. Future studies might consider other acquisition durations and applied activities; a prospective study evaluating the non-inferiority of low-dose acquisitions is ongoing (NCT05496920). Future prospective studies might test any potential clinical advantage, including at the patient-outcome level to determine the clinical utility of LAFOV systems in HNC. Although our study was adequately powered to test a set of pre-defined statistical hypotheses, we recognise that 64 patients cannot test all circumstances where high-count acquisitions might be of additional utility, such as in the detection of systemic metastases, for which larger cohorts will be needed.

This study demonstrated the clinical acceptability of 4 min acquisitions compared to 10-min acquisitions for the detection of LN metastases in HNC. Ten-minute acquisitions were associated with small and borderline-significant improvements in lesion detection and accuracy. We posit that the longer duration acquisitions which LAFOV systems afford might translate into improved diagnostic performance, which future studies might investigate.