Staging the axilla in breast cancer patients with ¹⁸F-FDG PET: how small are the metastases that we can detect with new generation clinical PET systems?

Abstract

Purpose: Point spread function (PSF) reconstruction improves spatial resolution throughout the entire field of view of a PET system and can detect smaller metastatic deposits than conventional algorithms such as OSEM. We assessed the impact of PSF reconstruction on quantitative values and diagnostic accuracy for axillary staging of breast cancer patients, compared with an OSEM reconstruction, with emphasis on the size of nodal metastases.

Methods: This was a prospective study in a single referral centre in which 50 patients underwent an (18)F-FDG PET examination before axillary lymph node dissection. PET data were reconstructed with an OSEM algorithm and PSF reconstruction, analysed blindly and validated by a pathologist who measured the largest nodal metastasis per axilla. This size was used to evaluate PET diagnostic performance.

Results: On pathology, 34 patients (68%) had nodal involvement. Overall, the median size of the largest nodal metastasis per axilla was 7 mm (range 0.5 - 40 mm). PSF reconstruction detected more involved nodes than OSEM reconstruction (p = 0.003). The mean PSF to OSEM SUVmax ratio was 1.66 (95 % CI 1.01 - 2.32). The sensitivities of PSF and OSEM reconstructions were, respectively, 96% and 92% in patients with a largest nodal metastasis of >7 mm, 60% and 40% in patients with a largest nodal metastasis of ≤7 mm, and 92% and 69% in patients with a primary tumour ≤30 mm. Biggerstaff graphical comparison showed that globally PSF reconstruction was superior to OSEM reconstruction. The median sizes of the largest nodal metastasis in patients with nodal involvement not detected by either PSF or OSEM reconstruction, detected by PSF but not by OSEM reconstruction and detected by both reconstructions were 3, 6 and 16 mm (p = 0.0064) respectively. In patients with nodal involvement detected by PSF reconstruction but not by OSEM reconstruction, the smallest detectable metastasis was 1.8 mm.

Conclusion: As a result of better activity recovery, PET with PSF reconstruction performed better than PET with OSEM reconstruction in detecting nodal metastases ≤7 mm. However, its sensitivity is still insufficient for it to replace surgical approaches for axillary staging. PET with PSF reconstruction could be used to perform sentinel node biopsy more safely in patients with a primary tumour ≤30 mm and with unremarkable PET results in the axilla.

Fig. 1

Impact of PSF reconstruction on quantitative values and number of detected nodes. Bland and Altman analysis for SUV_max (a) and node/background ratio (b) and the numbers of involved nodes accurately detected by PET with PSF reconstruction and OSEM reconstruction (c) are shown

Fig. 2

Flow-chart of clinical, pathological and PET status in 50 patients analysed (PSF-/PSF+ negative/positive by PSF reconstruction, OSEM-/OSEM+) negative/positive by OSEM reconstruction

Fig. 3

A 64-year-old patient with grade II infiltrating ductal carcinoma initially classified as T2N0c, in whom PSF reconstruction was positive and OSEM reconstruction was negative. Pathology revealed 4-mm and 6-mm lymph node metastasesred arrows. Both OSEM and PSF reconstructions are scaled to the same maximum value (a CT images, b pathological specimen (HES staining), c corresponding PET slices)

Fig. 4

A 51-year-old patient with grade III infiltrating ductal carcinoma initially classified as T3N0c, in whom both PSF reconstruction and OSEM reconstruction were positive, but PSF reconstruction depicted an additional involved lymph node (yellow arrows) (a CT images, b PSF reconstruction, c OSEM reconstruction). Both OSEM and PSF reconstructions are scaled to the same maximum value. Note how PSF reconstruction improves activity recovery in small lesions (red and yellow arrows) as compared with the largest node (orange arrows)

Fig. 5

Graphical comparison between the two algorithms. Each subgroup is defined according to the size of the primary tumour (a ≤30 mm, b >30 mm; 23 and 27 patients, respectively). Compared with OSEM reconstruction, a diagnostic procedure in the first quadrant (I) will be interpreted as superior overall, in the second quadrant (II) as superior for confirming the absence of metastases, in the third quadrant (III) as superior for confirming the presence of metastases, or in the last quadrant (IV) as inferior overall

Fig. 6

Sizes of the largest nodal metastasis per axilla in relation to nodal involvement status as detected by PET using PSF or OSEM reconstruction: PSF-/OSEM- patients with nodal involvement not detected by either reconstruction, PSF+/OSEM- patients in whom nodal involvement was detected only by PSF reconstruction, PSF+/OSEM+ patients in whom nodal involvement was correctly detected by both reconstructions). The extreme values and quartiles (when n >3) are shown together with the median values (longest bars) and mean values (shortest bars). The sizes were compared using the Kruskal-Wallis rank sum test