Prognostic value of [¹⁸F]FDG PET/CT and PERCIST criteria in patients with HER2-low metastatic breast cancer treated with trastuzumab deruxtecan

Abstract

Introduction

Trastuzumab deruxtecan (T-DXd) has recently reshaped the management of HER2-low metastatic breast cancer. While pivotal trials relied on CT-based RECIST 1.1, [¹⁸F]FDG PET/CT provides high diagnostic sensitivity and valuable semi-quantitative metrics, though its role in monitoring antibody–drug conjugates remains insufficiently defined. This study assessed the prognostic value of PET-derived parameters and PERCIST criteria in patients treated with T-DXd.

Methods

This retrospective study included all HER2-low metastatic breast cancer patients treated with T-DXd between 2022 and 2025 at the Strasbourg Europe Cancer Institute (ICANS) who underwent [¹⁸F]FDG PET/CT. Semi-quantitative parameters (SUVmax, SULpeak) and metabolic response according to PERCIST 1.0 were collected. Their association with overall survival (OS), metabolic progression-free survival (mPFS) and time-to-treatment change (TTT) was analyzed using Cox models and Kaplan–Meier curves.

Results

Forty-three patients were included. Baseline SULpeak of the most hypermetabolic lesion was an independent prognostic factor for OS, mPFS and TTT (p < 0.05). A ROC-derived cutoff of 6 identified two prognostic groups, with SULpeak < 6 associated with significantly improved outcomes. Early metabolic response was also linked to longer OS and PFS (p< 0.05), with median PFS of 9.2 vs. 4.2 months (HR 2.67; 95% CI 1.30–5.47). Combining baseline SULpeak and metabolic response (complete or partial metabolic response) defined three risk groups with significantly different OS and PFS.

Conclusion

[¹⁸F]FDG PET/CT provides significant prognostic information in HER2-low metastatic breast cancer treated with T-DXd. Baseline SULpeak and early PERCIST response enable meaningful risk stratification. The proposed prognostic model requires prospective validation before clinical implementation.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s00259-026-08029-4.

Introduction

Breast cancer remains the most prevalent malignancy among women worldwide, both in terms of incidence and mortality [1]. At the metastatic stage, disease management represents a major clinical challenge and requires careful optimization, given the limited availability of effective treatment options.

Trastuzumab deruxtecan (T-DXd), an anti-human epidermal growth factor receptor 2 (HER2) antibody–drug conjugate (ADC), is reshaping the therapeutic landscape for metastatic HER2-positive breast cancer. It is currently established as the standard of care in the second-line setting, as demonstrated in the DESTINY-Breast03 trial [2]. Moreover, T-DXd has shown substantial clinical efficacy in tumors with low and ultra-low HER2 expression [3, 4].

Structurally, T-DXd consists of a humanized monoclonal antibody targeting the HER2 receptor conjugated, via a cleavable tetrapeptide linker, to deruxtecan, a potent topoisomerase I inhibitor. Upon binding to HER2 receptors on tumor cell membranes, the ADC-receptor complex is internalized, after which the linker is enzymatically cleaved within the lysosomal compartment. The released cytotoxic payload then diffuses into the nucleus, inducing DNA double-strand breaks and subsequent apoptotic cell death. Notably, due to its high membrane permeability, deruxtecan exerts a bystander effect, mediating cytotoxic activity in neighboring tumor cells regardless of HER2 expression levels [5].

This distinctive mechanism underlies the remarkable efficacy of T-DXd observed in HER2-low breast cancers. In the DESTINY-Breast04 trial, treatment with T-DXd resulted in a median progression-free survival of 9.9 months versus 5.1 months in the control arm (p < 0.001), and a median overall survival of 23.4 months versus 16.8 months (p = 0.001) [4]. Response assessments in the pivotal trials leading to T-DXd approval were based on contrast-enhanced thoraco-abdomino-pelvic computed tomography (CT) scans evaluated according to RECIST 1.1 criteria.

Fluorodeoxyglucose positron emission tomography combined with computed tomography ([¹⁸F]FDG PET/CT) plays a key role in oncology for diagnosis, staging, and therapeutic monitoring, with diagnostic performance superior to that of CT alone. In the context of HER2-positive breast cancer, [¹⁸F]FDG PET/CT has been shown to be faster and more sensitive in detecting metastatic disease progression among patients treated with trastuzumab and pertuzumab or trastuzumab and lapatinib [6, 7]. Furthermore, application of PERCIST 1.0 criteria enables the identification of measurable lesions more frequently than RECIST 1.1, which often proves insufficient in cases of bone-only metastatic disease [8]. Beyond its diagnostic utility, [¹⁸F]FDG PET/CT also provides valuable prognostic information through semi-quantitative biomarkers. Multiple studies have demonstrated that elevated SUVmax values or high metabolic tumor volume (MTV) are associated with shorter progression-free and overall survival, including in metastatic breast cancers [9, 10].

Nevertheless, current guidelines remain ambiguous regarding the clinical application of PET/CT in breast cancer management, still considering it as an alternative to contrast-enhanced CT combined with bone scintigraphy for patient follow-up [11]. Moreover, data on its role in monitoring emerging antibody–drug conjugate (ADC) therapies, such as T-DXd, remain limited.

In this context, we aimed to determine the prognostic role of [¹⁸F]FDG PET/CT in patients with HER2-low metastatic breast cancer treated with T-DXd. The primary objective of this study was to evaluate the correlation between semi-quantitative PET parameters and survival outcomes.

Materials and methods

Study design and population

This retrospective, observational study was conducted at the Institut de Cancérologie Strasbourg Europe (ICANS). Eligible patients were female patients with histologically confirmed HER2-low breast cancer (defined according to the “Groupe d’Étude des Facteurs Pronostiques Immunohistochimiques dans les Cancers du Sein (GEFPICS)” recommendations as an immunohistochemistry (IHC) score of 1 + or 2 + without amplification on silver in situ hybridization or fluorescence in situ hybridization (SISH/FISH [12])).

To be included, patients were required to have received at least one cycle of T-DXd and to have undergone a baseline [¹⁸F]FDG PET/CT scan prior to treatment initiation, as well as at least one follow-up PET/CT scan during treatment. Metabolic reassessments were performed every 6 to 12 weeks throughout treatment.

PET/CT acquisition and reconstruction protocol

In accordance with standard practice guidelines, all patients were required to fast for at least six hours prior to undergoing [¹⁸F]FDG PET/CT imaging. An intravenous dose of [¹⁸F]FDG was administered at 3 MBq/kg body weight, followed by a resting period of 60 min. Whole-body acquisitions were performed from the vertex to mid-thigh. Imaging was conducted using two PET/CT systems: the Siemens Biograph Vision 600 and the Siemens Biograph mCT. Low-dose CT was performed for attenuation correction and anatomic localization. All images were independently reviewed by two experienced nuclear medicine physicians, with discrepancies resolved by consensus.

PET images were reconstructed using routine clinical ordered-subsets expectation maximization (OSEM) algorithms with time-of-flight and point-spread-function modelling on both systems: 3 iterations/5 subsets with a 3-mm Gaussian post-reconstruction filter for the Biograph Vision 600, and 2 iterations/21 subsets with a 3.5-mm Gaussian post-reconstruction filter for the Biograph mCT. These routine reconstruction protocols were obtained through a quantitative performance harmonization procedure performed at scanner installation using an [18 F]FDG-filled IEC body phantom, with the aim of providing comparable PET quantification between both systems. A phantom-based verification showed a maximum inter-scanner difference of 4.5% in 3D recovery coefficients; details are provided in Supplemental File 1 and Supplemental Fig. 1.

Both PET/CT systems were operated within the same institution using standardized acquisition protocols including harmonized injected activity and uptake time in accordance with current EANM guidelines. Routine quality control and calibration procedures were performed for both systems.

Measured parameters and collected data

Demographic, clinical, and histological data (including age at diagnosis, age at metastatic diagnosis, body weight, ECOG performance status, HER2 expression, and hormone receptor (HR) status were collected for all patients. Data on systemic treatment history—such as initiation dates, therapeutic modalities, and the number of treatment lines received—were also recorded.

For each [¹⁸F]FDG PET/CT examination, the number and distribution of metastatic sites were manually recorded, and the most metabolically active target lesion was identified. Target lesions were defined in accordance with PERCIST 1.0 criteria [13] as having a SULpeak greater than 1.5 × mean hepatic SUL + 2 SD, with a longest diameter > 1.5 cm (or > 1.0 cm for lymph nodes). The mean hepatic SUL was derived from a 3-cm spherical VOI positioned within the right hepatic lobe, or from aortic activity in cases of extensive hepatic tumor involvement. The SUVmax, SULmean, and SULpeak of the most hypermetabolic lesion were measured, with the target lesion permitted to differ between scans. Those data were extracted using the Syngo.via software version VB30B.

Metabolic treatment response was assessed following PERCIST 1.0 criteria and categorized as complete (CMR), partial metabolic response (PMR; >30% and > 0.8-unit decrease in SULpeak), progressive metabolic disease (PMD; >30% and > 0.8-unit increase or appearance of new lesions), or stable metabolic disease (SMD). A single-lesion approach was applied for PERCIST assessment, based on the most metabolically active lesion (highest SULpeak), in accordance with recommended practice. RECIST 1.1-based response assessments were not systematically available for all patients and were therefore not included in the present analysis.

The clinical outcomes of the study included the objective response rate (ORR) and metabolic progression-free survival (mPFS). ORR was defined as the proportion of patients who achieved either a CMR or PMR at any time during therapy, whereas patients exhibiting PMD or SMD were categorized as non-responders. mPFS was calculated as the interval between initiation of T-DXd and the occurrence of metabolic progression according to PERCIST criteria. The time to treatment change (TTT) was defined as the interval between initiation of T-DXd and any treatment modification or discontinuation, regardless of the underlying reason (disease progression, toxicity, or physician decision).

Analysis and statistical methods

A univariate Cox proportional hazards regression analysis was performed to identify factors potentially associated with survival and time to treatment change. Given the limited sample size and number of events, particular attention was paid to model parsimony. The number of variables included in multivariable Cox regression models was deliberately restricted based on clinical relevance and preselection in univariate analysis (p < 0.2), in order to reduce the risk of overfitting. All statistical tests were two-sided, and p-values < 0.05 were considered statistically significant.

Survival distributions were estimated using the Kaplan–Meier method and comparison between groups were conducted using the log-rank test. Hazard ratios (HR) and their corresponding 95% confidence intervals (CI) were calculated for each variable. Statistical analyses were performed using MedCalc software (version 23.0.8) and GraphPad Prism (version 9.0.0).

To account for the potential immortal time bias associated with early metabolic response assessment, survival analyses based on PERCIST response were conducted using a landmark approach. Only patients who were alive and evaluable at the time of the first follow-up PET/CT (approximately 3 months after treatment initiation) were included in these analyses. Patients who experienced disease progression or death before this time point were not included in response-based subgroup analyses.

Results

Patients’ characteristics

A total of 43 female patients were included in the analysis. Table 1 summarizes the patients’ characteristics. PET/CT imaging was performed using a Siemens Biograph 600 scanner for 21 patients, and a Siemens Biograph mCT scanner for 22 patients. The median age at T-DXd initiation was 61 years, and 86% of patients had hormone receptor–positive (HR+) invasive carcinoma. Prior to initiation of T-DXd, 65% of patients had received four or more lines of systemic therapy (range: one to > six lines).

Table 1.

Summary of patients’ characteristics

Patients’ characteristics	Results
Demographics characteristics (years)	Median (range)
Age at metastatic diagnosis	58 (37–77)
Age at T-DXd initiation	61 (39–80)
Time to metastatic progression	1.6 (0–16.3)
ECOG performance status at T-DXd initiation	n (%)
0	3 (7%)
1	33 (77%)
2	7 (16%)
Number of prior systemic treatment
< 4	15 (35%)
≥ 4	28 (65%)
Histology
Invasive ductal carcinoma (IDC)	40 (93%)
Invasive lobular carcinoma (ILC)	3 (7%)
Hormone receptor (HR) status
ER+/PR+	31 (72%)
ER+/PR -	6 (14%)
ER-/PR +	0 (0%)
ER-/PR -	6 (14%)
HER2 status
HER2 2 + without amplification	25 (58%)
HER2 1+	18 (42%)
Number of metastases on baseline PET/CT
0–4	11 (26%)
5–9	6 (14%)
≥ 10	22 (51%)
Metastases topography
Exclusive bone	11 (26%)
Visceral +/- bone	32 (74%)
Liver	18 (42%)
Lung	7 (16%)
Brain	1 (2%)
Lymph nodes	16 (37%)
Peritoneal	3 (7%)
Other (local, muscles, pleural)	8 (19%)
Baseline PET biomarkers of the target lesion	Median (IQR)
SUVmax	10.8 (4.2–44.5)
SULmean	2.2 (1.0–7.9)
SULpeak	4.9 (2.3–16)
PERCIST response on 1 st reassessment PET	n (%)
CMR	7 (16%)
PMR	10 (23%)
SMD	9 (21%)
PMD	17 (40%)

ER: Estrogen receptor/PR: progesterone receptor/CMR: complete metabolic response/PMR: partial metabolic response/SMD: stable metabolic disease/PMD: progressive metabolic disease/Time to metastatic progression was defined as the interval between initial breast cancer diagnosis and the diagnosis of metastatic disease

The median time between baseline PET/CT and first dose of T-DXd was 17 days. Most patients presented with extensive metastatic burden, with 65% presenting more than 5 metastatic lesions and 74% showing visceral involvement. At baseline PET imaging, metastatic lesions were as follows: bone, liver, lymph node, and lung. The median SUVmax of the target lesion was 10.8; and the median SULpeak was 4.9.

The median time to the first metabolic response assessment by PET/CT was 2.9 months (IQR: 2.5–4.2). On this initial reassessment, 39% of patients demonstrated a metabolic response (ORR), with 16% achieving a complete metabolic response and 23% a partial metabolic response. Survival analyses according to early metabolic response were restricted to patients evaluable at the time of first PET reassessment, in accordance with the predefined landmark approach. After a median follow-up of 12.5 months, the median metabolic progression-free survival was 6.18 months. At the time of analysis, 15 patients had died during follow-up.

Association between baseline PET biomarkers and clinical outcomes

In univariate Cox regression analysis, both the SULpeak and SUVmax of the most hypermetabolic lesion on baseline PET were significantly associated with metabolic progression-free survival and time to treatment change (p < 0.05).

The variables entered into the multivariable models included baseline SULpeak, PERCIST response, hormone receptor status, delta SUL, and number of metastatic lesions. After adjustment, baseline SULpeak remained independently associated with OS, mPFS and time to treatment change (p = 0.0067 for OS; p = 0.003 for mPFS and p = 0.02 for time to treatment change, respectively).

Receiver operating characteristic curve (ROC) analysis yielded an area under the curve of 0.869 (95% CI, 0.692–0.965; p = 0.0006). The optimal cutoff identified using the Youden index was 6.02 for discriminating between two prognostic subgroups, corresponding to a sensitivity of 87.5% and a specificity of 95.2%. This threshold was subsequently applied to survival analyses, and as shown in Fig. 1, patients with SULpeak ≥ 6 (Group B) had significantly shorter OS (Fig. 1A) and mPFS (Fig. 1B) compared with those with SULpeak < 6 (Group A) (median OS, 9.6 months vs. not reached; HR 8.1; 95% CI 2.2–29.3; p = 0.0014; and median mPFS, 3.8 vs. 10.8 months; HR 8.7; 95% CI 3.3–22.4; p < 0.0001). Similarly, time to treatment change was shorter in patients with SULpeak ≥ 6 compared with those with lower SULpeak values (4.4 vs. 11 months; HR 4.6; 95% CI 1.9–11.0; p = 0.0005) (Fig. 1C).

Fig. 1.

Kaplan–Meier survival curves depicting overall survival (OS), metabolic progression-free survival (mPFS), and time to treatment change according to baseline SULpeak values and metabolic response assessed by PERCIST 1.0 criteria. Panels A–C illustrate OS (A), mPFS (B), and time to treatment change (C) stratified by baseline SULpeak of the most hypermetabolic lesion on PET. Group A (blue curves) included patients with baseline SULpeak < 6, whereas Group B (green curves) comprised patients with baseline SULpeak ≥ 6. Panels D–F show OS (D), mPFS (E), and time to treatment change (F) according to metabolic response at 3 months, assessed on the first follow-up PET scan using PERCIST 1.0 criteria. Responders (Group 0, blue curves) demonstrated significantly improved survival outcomes compared with non-responders (Group 1, green curves), including longer OS and mPFS, with a similar result observed for time to treatment change

Association between PERCIST response at 3 month and clinical outcomes

In both univariate and multivariate Cox regression analyses, the objective response rate assessed on the first follow-up PET according to PERCIST 1.0 criteria was significantly associated with OS and mPFS (p = 0.0145 and p = 0.0304, respectively, in multivariate analysis). A nonsignificant trend toward association was also observed for time to treatment change (p = 0.09).

Patients classified as metabolic responders exhibited markedly improved survival outcomes compared with non-responders with a median OS not reached versus 15 months (HR 5.02; 95% CI 1.80–14.04; p = 0.021; Fig. 1D) and median mPFS of 9.2 versus 4.2 months (HR 2.67; 95% CI 1.30–5.47; p = 0.0073; Fig. 1E). Similar results were observed for time to treatment change (HR 2.43; 95% CI 1.22–4.82; p = 0.0113; Fig. 1F).

Risk stratification based on baseline SULpeak and 3-month PERCIST response

Given the significant association between the baseline SULpeak of the most hypermetabolic lesion (threshold: 6) and a metabolic response with survival outcomes, patients were stratified into three distinct prognostic groups:

• Low-risk group: baseline SULpeak < 6 and a metabolic response (CMR or PMR) on the first follow-up PET; n = 12 (28%).

• Intermediate-risk group: presence of a single adverse prognostic feature, either baseline SULpeak > 6 or absence of a metabolic response on the first follow-up PET; n = 20 (46%).

• High-risk group: both adverse features present—baseline SULpeak > 6 and absence of a metabolic response (SMD or PMD) on the first follow-up PET; n = 11 (26%).

Median OS differed across groups, it was not reached in the low-risk group, compared with 23.5 and 8.0 months in the intermediate- and high-risk groups, respectively (Fig. 2A). Similarly, median mPFS was not reached, 8.3, and 3 months for the low-, intermediate-, and high-risk groups, respectively (Fig. 2B). The median time to treatment modification was shortest in the high-risk group (4.1 months), compared with 9.2 and 18.7 months in the intermediate- and low-risk groups, respectively (Fig. 2C).

Fig. 2.

Kaplan–Meier curves for OS (A), mPFS (B), and time to treatment change (C) according to prognostic group. The blue curve represents Group 0 (low-risk), the green curve represents Group 1 (intermediate-risk), and the orange curve represents Group 2 (high-risk)

No significant differences in OS or time to treatment modification were observed between the low- and intermediate-risk groups. In contrast, the high-risk group exhibited significant differences compared with both other groups (for OS, HR 21.1; 95% CI 4.3–104.4; p = 0.0002 for low-risk group and HR 61.8; 95% CI 10.8–351; p < 0.0001; Fig. 2A).

Regarding metabolic PFS, patients in the low-risk group exhibited a marked survival advantage relative to the high-risk group (HR 23.9; 95% CI 6.6–86.3; p < 0.0001) and to the intermediate-risk group (HR 2.5; 95% CI 1.0–6.3; p = 0.0415). Moreover, the intermediate-risk group showed a significant benefit compared with the high-risk group (HR 22.4; 95% CI 6.4–77.5; p < 0.0001) (Fig. 2B).

Figure 3 presents two representative cases in which our stratification framework was applied, illustrating the typical clinical trajectories observed in each prognostic group and thereby reinforcing the validity of this approach. The low-risk patient presented with hepatic, bone, and interbronchial metastases, with the most hypermetabolic lesion located in an interbronchial lymph node (SULpeak = 4.1). The first follow-up imaging – 1.8 months after introduction of T-DXd – demonstrated a partial metabolic response, with no evidence of disease progression or death throughout follow-up of 16.7 months (date of censoring). In contrast, the high-risk patient presented with pulmonary and bone metastases, with the most hypermetabolic lesion located in the left scapular bone (SULpeak = 10.0). Follow-up imaging revealed progressive metabolic disease 3 months after T-DXd introduction, with emergence of new lung lesions. The patient died 5.7 months after the therapy initiation.

Fig. 3.

Baseline and follow-up PET MIP images of a low-risk patient (A) and high-risk patient (B) according to our classification. Orange arrows indicate baseline lesions and their evolution; green arrows highlight newly appearing lesions

Discussion

In this retrospective study, we investigated the prognostic value of baseline [¹⁸F]FDG PET/CT parameters and early metabolic response according to PERCIST 1.0 in patients with HER2-low metastatic breast cancer treated with T-DXd. Our findings demonstrate that both baseline SULpeak and early metabolic response at three months are independent predictors of overall survival, metabolic progression-free survival, and time to treatment modification. These results support the role of PET-based biomarkers as relevant tools to refine clinical decision-making in a therapeutic context where validated prognostic indicators remain scarce.

The prognostic model generated by combining baseline SULpeak with early metabolic response enabled a robust stratification into low-, intermediate-, and high-risk categories. Patients exhibiting both an elevated SULpeak (≥ 6) and lack of early metabolic response demonstrated markedly inferior clinical outcomes, whereas those with a low SULpeak and evidence of early response experienced substantially improved survival. This dual-parameter approach is clinically attractive, as it relies on two easily obtainable, reproducible PET metrics and can be implemented in routine practice without additional imaging complexity.

Our results are consistent with prior reports demonstrating the prognostic utility of semi-quantitative PET parameters, including SUVmax, metabolic tumor volume, and heterogeneity indices, in metastatic breast cancer [14]. A French multicentric analysis evaluating antibody–drug conjugates similarly highlighted the predictive value of metabolic tumor burden markers such as TMTV and Dmax [15]. In this context, SULpeak represents an appealing biomarker: unlike SUVmax, it is less affected by statistical noise and reconstruction artifacts, and unlike volumetric measurements, it does not require specialized segmentation software. Its prognostic relevance has also been validated across multiple solid malignancies [16, 17], reinforcing its robustness as a quantitative PET metric. From a biological perspective, increased FDG uptake reflects enhanced tumor glycolytic activity, and has been associated with higher proliferative rates, aggressive molecular subtypes, and poorer clinical outcomes in breast cancer [18]. In the context of antibody–drug conjugates such as T-DXd, elevated metabolic activity may also reflect greater tumor heterogeneity and clonal complexity, potentially influencing treatment sensitivity and durability of response. In addition, HER2-low breast cancer represents a biologically heterogeneous entity, encompassing subgroups with variable aggressiveness and treatment response profiles [19]. This may partly explain why patients with higher baseline SULpeak exhibit poorer survival outcomes despite initial treatment response, reflecting the complex interplay between tumor metabolism, biological aggressiveness, and therapeutic sensitivity.

The early timing of metabolic assessment in our cohort (median 2.9 months) further supports the use of PET/CT for early treatment monitoring under T-DXd. Comparable findings have been reported in other settings, including endocrine-based therapies and cytotoxic chemotherapy. Sirico et al. [20] showed that in ER+/HER2 − breast cancer patients treated with everolimus and exemestane, a three-month metabolic response (defined as > 28% reduction in SUVmax) was significantly associated with longer PFS. Similarly, Kitajima et al. [21] reported that among patients—39% of whom were HER2-positive—those achieving a metabolic response (CMR or PMR) after a single chemotherapy cycle exhibited a significantly longer PFS and a trend toward improved OS. Given the relatively high FDG avidity of HER2-low tumors observed in our study, the incremental value of PET may be even greater in this subgroup than in HER2-negative disease, where PET performance is typically less robust. Notably, a substantial proportion of patients (40%) exhibited early metabolic progression at first evaluation. This likely reflects the advanced and heavily pretreated nature of our cohort, as well as the biological heterogeneity of HER2-low metastatic breast cancer, which includes subgroups with more aggressive clinical behavior, particularly in the metastatic setting.

From a clinical perspective, early metabolic response assessment using [¹⁸F]FDG PET/CT may provide valuable insight into treatment effectiveness in patients receiving trastuzumab deruxtecan. In particular, the identification of patients with early metabolic progression could help to recognize those at higher risk of poor outcomes. In selected situations, this information may support clinical decision-making, for example by prompting closer monitoring or discussion of alternative therapeutic strategies. However, given the retrospective design and limited sample size, these findings should not be used alone to guide treatment modification in routine clinical practice and require prospective validation.

The time to treatment change represents a composite endpoint that may be influenced by factors other than objective disease progression, including treatment tolerance, physician decision-making, and patient preference. As such, it may introduce a degree of subjectivity and should be interpreted with caution. Nevertheless, TTT reflects real-world clinical practice and may provide complementary information on treatment dynamics, particularly in settings where early treatment adaptation is common. Importantly, the consistency of the associations observed across TTT, metabolic progression-free survival, and overall survival supports the clinical relevance of the identified prognostic factors.

The application of PERCIST 1.0 criteria proved feasible and clinically informative in our real-world setting. We elected to use a single-lesion analysis strategy, which enhances reproducibility and reduces inter-observer variability [22]. Consequently, the mPFS endpoint used in this study reflects disease evolution according to a PERCIST framework based on the most metabolically active lesion, including the appearance of new lesions, rather than a comprehensive assessment of whole-body metabolic tumor burden. This distinction should be considered when interpreting the prognostic value of mPFS. Prior analyses have shown that such a simplification does not compromise prognostic accuracy, particularly for therapies with diffuse cytotoxic effects such as ADCs. Notably, the threshold SULpeak value identified in our cohort (6.0) emerged from ROC analysis and may not be directly generalizable to other centers, given differences in scanner types, acquisition protocols, and reconstruction parameters. Nonetheless, the concept of deriving a center-specific SULpeak cutoff appears relevant for harmonizing PET-based prognostic stratification across institutions.

Despite similarities in patient characteristics, the median PFS observed in our study was shorter than that reported in DESTINY-Breast04. Several factors may account for this difference, including heavier pretreatment, real-world imaging intervals, and the use of PET-based endpoints, which may be more sensitive to early progression than CT alone. A direct comparison between PET-based and CT-based response assessment (RECIST 1.1) was not feasible in this study, as standardized RECIST evaluations were not consistently available due to the retrospective design. Accordingly, the present study was not designed to assess the incremental value of PET/CT over contrast-enhanced CT, but rather to evaluate the prognostic relevance of PET-derived biomarkers in this specific clinical setting. Nevertheless, emerging evidence suggests that PET identifies progression earlier than CT; consequently, while this may enhance prognostic accuracy, it may also shorten measured PFS relative to CT-based trials. In particular, the prospective study by Vogsen et al. [8] showed that [¹⁸F]FDG PET/CT could distinguish responders from non-responders approximately six months earlier than CT alone, and was significantly associated with PFS. Consistently, a subsequent prospective study from the same group demonstrated that PET-based response assessment was more strongly associated with both progression-free and disease-specific survival than conventional CT [23]. Taken together, these findings reinforce the clinical reliability of PET/CT for the early detection of disease progression.

The clinical implications of early metabolic progression warrant careful consideration. Although early PET progression was associated with poorer outcomes, immediate treatment modification in a context of limited post-T-DXd options may prematurely exhaust available therapies. In our cohort, treatment was maintained in 10 patients despite early metabolic progression, including six without clinical deterioration; however, most subsequently showed further progression, while only one patient remained metabolically stable. The remaining PMD cases corresponded to oligoprogressive disease treated locally. These findings suggest that delaying treatment modification may have limited benefit in some patients, although the optimal management of early PET-detected progression remains uncertain.

This study has several limitations. Its retrospective, single-center design and relatively small sample size, with a limited number of events, may reduce statistical power and increase the risk of overfitting in multivariable analyses, as reflected by the relatively wide confidence intervals. Second, the ROC-derived SULpeak cutoff of 6.02 was not validated in an independent internal or external cohort and should therefore be considered exploratory. Although acquisition and reconstruction protocols were harmonized within our institution, overfitting cannot be excluded. The generalizability of this threshold across other centers, scanner platforms, and reconstruction protocols remains uncertain and requires dedicated external validation. Third, RECIST 1.1-based assessments were not systematically available, precluding direct comparison between PET- and CT-based response evaluation and limiting conclusions regarding the incremental value of PET/CT. Fourth, the timing of the first follow-up PET/CT varied across patients, which may have influenced the assessment of early metabolic response due to differences in treatment exposure. Although the median time to evaluation was relatively short, this variability may introduce heterogeneity in response classification. Fifth, PET-based inclusion criteria may have excluded patients with non-FDG-avid or sub-centimetric lesions, potentially limiting the generalizability of the findings. Finally, despite the use of a landmark approach to reduce immortal time bias, the evaluation of early metabolic response remains subject to inherent limitations of retrospective analyses.

Accordingly, these findings should be considered hypothesis-generating and require confirmation in larger prospective, multicenter studies with standardized imaging protocols and external validation. In addition, no formal internal validation procedure (bootstrap resampling or cross-validation) was performed, which further limits confidence in the generalizability and stability of the proposed prognostic model.

Nevertheless, the study reflects real-world practice, including heavily pretreated patients, and provides clinically relevant evidence supporting the use of PET for prognostic refinement in HER2-low metastatic breast cancer. Prospective trials directly comparing PET/CT with contrast-enhanced CT in the monitoring of T-DXd—such as the PERPETUAL study (NCT07130344)—are needed to confirm the prognostic thresholds identified and to clarify the optimal integration of metabolic imaging within treatment algorithms. Ultimately, studies evaluating PET-guided treatment strategies based on early metabolic response will be required to determine whether PET-driven management can improve patient outcomes.

Conclusion

This study highlights the prognostic value of [¹⁸F]FDG PET/CT in patients with HER2-low metastatic breast cancer treated with Trastuzumab Deruxtecan—both at baseline, through assessment of SULpeak, and during treatment monitoring, according to PERCIST 1.0 criteria.

Our findings emphasize the potential of metabolic-based prognostic stratification to inform and optimize therapeutic decision-making, thereby reinforcing the expanding role of nuclear medicine within a personalized approach to metastatic breast cancer management. The proposed risk stratification model should be considered exploratory and requires prospective validation before being incorporated into routine clinical decision-making.

Prospective studies with larger cohorts are warranted to validate these results and to directly compare the clinical utility of PET/CT and contrast-enhanced CT in the longitudinal evaluation of patients receiving T-DXd.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

The study’s design, material preparation, data collection and analysis was performed by JF, FS and CB. JS performed the harmonization procedure of PET/CT devices. AI-assisted language editing (limited to grammar and clarity improvements) was performed using ChatGPT (OpenAI). The first draft of the manuscript was written by JF and FS, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Data Availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval

This study was performed in line with the principles of the Declaration of Helsinki and was approved by local Institutional Review Board (Faculty of Medicine/Strasbourg University Hospitals, IRB 2024-22).

Consent to participate

Informed consent was obtained from all individual participants included in the study.

Consent to publish

The authors affirm that human research participants provided informed consent for publication of the images in Figure 3.

Competing Interests

The authors have no relevant financial or non-financial interests to disclose.