Real-time quantification during indocyanine green fluorescent-guided surgery in canine soft tissue sarcomas and mast cell tumors

Abstract

The aim of this pilot study was to evaluate the feasibility and utility of real-time tumor fluorescence-guided surgery (TFGS) with near-infrared fluorescence indocyanine green (NIRF-ICG) for the removal of spontaneous soft tissue sarcoma (STS) and mast cell tumor (MCT) at first presentation or recurrence in dogs. A dose of 0.5 mg/kg of ICG was administered 24 h before surgery, with a tumor-to-background (TBR) and signal-to-background ratio (SBR) cut-off of 2. Twenty-four dogs with 28 tumors (14 MCTs and 14 STSs) were included. TFGS was feasible in all STSs and in 11/14 MCTs. The median TBR of STS and MCT was 13 (IQR 26–9.4) and 5.8 (IQR 19–3.6) respectively. Considering the residual fluorescence in the wound bed and the histopathological results the sensitivity and specificity of NIRF-ICG excision were 80% and 78% in STS and 60% and 50% in MCT, respectively. Overall, resection was extended during TFGS in 64% of STSs and 55% of MCTs. These results convey the feasibility of TFGS with NIRF-ICG in canine STS surgical extirpation. However, the findings indicate more limited performance in MCT. The different performance in STS and MCT needs to be investigated in a larger population.

Introduction

Achieving histologically tumor-free surgical margins is essential for the successful treatment of solid malignant tumors^1–3. Despite improvements in the planning of surgery for canine soft tissue sarcomas (STS) and mast cell tumors (MCT) over the last decade^4,5 and updates in decision-making regarding radical versus wide versus marginal excision based on tumor size and location^2,3,6–10, accurately identifying tumor boundaries during surgery remains challenging. The need for the development of tools that assist surgeons during tumor excision and enhance surgical precision through real-time intraoperative assessment of the distinction between neoplastic and non-neoplastic tissue represents a new frontier in canine surgical oncology^4,11,12.

In human surgical oncology, tumor fluorescence-guided surgery (TFGS) using near-infrared fluorescence (NIRF) has emerged as a valuable intraoperative technique to differentiate neoplastic from non-neoplastic tissue and optimize resection^13–16. Indocyanine green (ICG) is a non-target fluorophore that, after a defined interval from intravenous injection, shows passive accumulation in neoplastic tissue (allowing the so-called “second window’’ ICG)^17,18. Although this phenomenon is not yet fully understood, most studies describe it as being due to the enhanced permeability and retention (EPR) effect, which results from aberrant tumor vascularization¹⁷. More recent research also reports tumor-specific uptake via clathrin-mediated endocytosis¹⁹. Taking advantage of these ‘’second window’’ phenomena, promising results have been reported by the application of NIRF-ICG in the surgical treatment of human soft tissue sarcomas and other solid tumors, with improved complete excision and reduced local recurrence rates^13,14,20.

Recently, research in small animal surgery has started exploring NIRF-ICG in different areas such as sentinel lymph node mapping, reconstructive surgery, and tissue perfusion assessment^21–24. However, despite the rising interest in NIRF-ICG, its use for TFGS is still not well established in small animals’ clinical practice and data on its application remain limited to translational medicine research^25–27, with some studies that have applied targeted fluorophores^11,28–31 or free-ICG (not conjugated to any specific molecular targeting agent)^25,26 for TFGS in canine STS and MCT. Given the paucity of research on TFGS, combined with the predominance of translational medicine studies on canine tumors, there is to date a significant heterogeneity of published data on ICG dosage (ranging from 0.5 to 7.5 mg/kg)^{18,19,24,26,32} and administration timing (ranging from 1 to 24 h before surgery)^14,20,25,30. Furthermore, fluorescence quantification, most expressed as tumor-to-background ratio (TBR) or signal-to-background ratio (SBR), shows considerable variability both in the devices used and in the cut-off values reported across veterinary and human literature. Reported TBR or SBR thresholds for distinguishing healthy from neoplastic tissue range from 1.2 to 3^{13,14,20,25,26,33,34}.

Hence, we designed a pilot study using a free-ICG dose of 0.5 mg/kg, an administration timing of 24 h before surgery, and a TBR and SBR with a cut-off of 2 to address the following three objectives:

• To evaluate the real-time intraoperative feasibility of TFGS with NIRF-ICG in spontaneous canine STS and MCT;

• To compare the real-time quantification of the wound bed fluorescence after excision with the histopathological margins status of the excised tumor;

• To explore real-time fluorescence quantification guidance during tumor resection and intraoperative decision-making.

Results

Study population

Among 30 animals evaluated for surgical purposes during the study period, 4 were excluded: 3 dogs had microscopic disease in the scar and 1 had local recurrence and distant metastasis. Twenty-six dogs met the inclusion criteria and were enrolled for the study. The most represented breed was the mixed breed (n = 8; 31%) followed by Labrador Retriever (n = 5; 19%) and Belgian Malinois (n = 2; 8%). The remaining 11 (42%) dogs each belonged to a different breed. Thirteen dogs (50%) were spayed female, 7 (27%) intact males, 3 (11%) intact females and 3 (12%) neutered males. Median age was 8.4 years (range 2.5–14.2 years. IQR: 11 − 6.3). The median bodyweight was 23 kg (IQR 31–10 kg) and 14 dogs (54%) were normal weight, while 12 (46%) were overweight.

Twenty-four dogs presented single tumor while 2 dogs presented simultaneous multiple tumors (1 dog had simultaneously 1 STS and 1 MCT, and 1 dog had 2 simultaneous MCTs). A total of 28 tumors were included (14 MCTs and 14 STSs). Recorded clinical variables of the tumors are reported in Table 1.

Table 1.

Tumors clinical variables.

Variables	STS (n = 14)		MCT (n = 14)
	N°	%	N°	%
Presentation First presentation Recurrence	11 3	79% 21%	12 2	86% 14%
Anatomical location Limb Head and Neck Inguinal/Perineal Trunk Digit	9 2 0 2 1	64% 14% 0% 14% 7%	5 3 2 4 0	36% 21% 14% 29% 0%
Ulceration No Yes	10 4	71% 29%	13 1	93% 7%
Clinical skin mobility overlying the tumor No Yes	6 8	43% 57%	7 7	50% 50%
Tumor clinically fixed to deeper fascia No Yes	10 4	71% 29%	13 1	93% 7%
Skin pigmentation No Yes	10 4	71% 29%	13 1	93% 7%
Tumor size (mm) Median IQ25% IQ75%	30 20 49		20 12 28

STS soft tissue sarcomas, MCT mast cell tumors.

The performed surgical dose was wide excision in 19/28 (68%) tumors (10 STSs and 9 MCTs) and a planned narrow excision in 9/28 (32%) tumors (4 STS and 5 MCT). The median surgical time for STSs and MCTs excision was respectively 50 (IQR 94–40) minutes and 40 (IQR 30–55) minutes. Post-operative surgical complications emerged in 7/28 (25%) surgical sites of which five mild complications-Grade 1 (seroma and hematoma) and two moderate complications-Grade 2 (dehiscence) that required a surgical revision.

TFGS and surgical procedures

No dogs showed adverse events during and after intravenous ICG administration.

Phase 1: tumor surgical resection

The technique resulted feasible in 14/14 (100%) STSs and in 11/14 (79%) MCTs. In 3/14 (21%) MCTs the TBR registered was < 2, making the fluorescence of the tumor undistinguishable from the background both with intact skin and after skin incision, therefore the technique was considered as failed. The 3 failed MCTs were excluded from the following analysis. Among the remaining cases, one subcutaneous MCT exhibited a TBR < 2 prior to incision; following skin incision and tumor exposure, the TBR increased to 14.

The median TBR of STS and MCT was 13 (IQR 26–9.4) and 5.8 (IQR 19–3.6) respectively. No statistical difference was present comparing populations of the 2 tumors in TBR values (p = 0.09).

During resection, the surgeon reported that the TFGS influenced the surgical procedure by increasing the planned resection of adjunctive tissue in 6/14 (43%) STSs and 4/11 (36%) MCTs. In all cases, the increase of lateral margins recorded (minimum of 0.5 cm to a maximum of 0.8 cm) did not change the planned surgical dose category. The adjunctive lateral tissue resection involved only one of the margins in all cases. No increase of deep margins was recorded in Phase 1.

Phase 2: wound bed fluorescence assessment before surgical defect reconstruction

The median SBR of STS and MCT was 1.1 (IQR 3.6–0.74) and 2 (IQR 5.4–1.3), respectively. No statistical difference was observed among the 2 tumor populations in SBR values (p = 0.3686). After tumor removal, 12 wound beds (6 STS and 6 MCT) had residual fluorescence. The qualitative and quantitative analysis of the fluorescence in the wound bed with NIRF-ICG after STS and MCT removal are reported in Table 2. Residual single-spot fluorescence was present in 9/12 wound beds (6 STSs and 3 MCTs): the surgeon performed adjunctive tissue resection in 8/9 wound beds (6/6 STSs and in 2/3 MCTs). This procedure on wound bed did not change the planned surgical dose category. In the remaining MCT case (carpal region), further resection was not feasible due to the proximity to functional neurovascular structures. Diffuse residual fluorescence was observed in 3/12 wound beds, but no biopsies or further resections were performed, due to the proximity to functional neurovascular structures.

Table 2.

Wound bed qualitative and quantitative analysis with NIRF-ICG in STS and MCT population.

Variables	STS (n = 14)		MCT (n = 11)
Residual fluorescence No Yes	8 6	57% 43%	5 6	45% 55%
Residual fluorescence distribution* Diffuse Single spot	0 6	0% 100%	3 3	50% 50%
SBR on the wound bed Median IQ25% IQ75%	1.1 0.74 3.6		2 1.3 5.4

STS soft tissue sarcomas, MCT mast cell tumors, SBR signal to background ratio, SD standard deviation, IQ Interquartile. *calculated in case of residual fluorescence.

The influence of the TFGS with NIRF-ICG on the intraoperative decision-making in Phase 1 and 2 is reported in Table 3. Overall, in 64% of STS and in 55% of MCT an increase of resection compared to the planned one was performed during the TFGS (Table 3).

Table 3.

Influence of the TFGS with NIRF-ICG on the surgery in STS and MCT population.

Variables	STS (n = 14)		MCT (n = 11)
Phase 1: increase resection during excision No Yes	8 6	57% 43%	7 4	64% 36%
Phase 2: increased wound bed tissue resection No Yes	8 6	57% 43%	9 2	82% 18%
Overall increased tissue resection (Phase 1 + Phase 2) No Yes	5 9	36% 64%	5 6	45% 55%

STS soft tissue sarcomas, MCT mast cell tumors.

All removed sentinel lymph nodes were radioactive according to lymphoscintigraphy and showed fluorescence uptake even if ICG was not used for SLN mapping.

Histopathological results

Detailed histopathology analyses of the 14 MCTs and 14 STSs are reported in Table 4. Overall, 10/28 tumors had infiltrated margins (5 STSs and 5 MCTs). Considering the 3 failed MCTs cases, they were all subcutaneous MCT (2 combined pattern and 1 infiltrative pattern) and one of them had infiltrated margins.

Table 4.

Histopathological characteristics of STS and MCT populations.

Tumor	Histological characteristics	N°	%
STS (n = 14)	Histotype Undifferentiated Grade III Perivascular wall tumor Grade I Grade II Grade III Fibrosarcoma Grade II Status of margins: Clean Infiltrated	2 5 2 2 3 9 5	14% 36% 14% 14% 22% 64% 36%
MCT (n = 14)	Cutaneous (n = 7) Cutaneous MCT Patnaik II/Kiupel low Status of margins: Clean Infiltrated	7 5 2	100% 71% 29%
	Subcutaneous (n = 7) Infiltrative pattern Combined pattern Status of margins: Clean Infiltrated	4 3 4 3	57% 43% 57% 43%

STS soft tissue sarcomas, MCT mast cell tumors.

Based on the presence of residual fluorescence and the results of histopathological margins status of resected tumors, the sensitivity and specificity of NIRF-ICG in STS excision were respectively 80% and 78% and in MCT were respectively 60% and 50%. Tables 5, 6 and 7.

Table 5.

Histological results and residual fluorescence in the wound bed.

Tumor		Infiltrated margins	Clean margins	Total
STS	Residual fluorescence	4 (True positive)	2 (False positive)	6
	No residual fluorescence	1 (False negative)	7 (True negative)	8
	Total	5	9	14
MCT	Residual fluorescence	3 (True positive)	3 (False positive)	6
	No residual fluorescence	2 (False negative)	3 (True negative)	5
	Total	5	6	11

Table 6.

Diagnostic performance indicators of NIRF-ICG in the wound bed analysis for STS and MCT with 95% Wilson Confidence Intervals.

Indicator	STS		MCT
	Result	95% Confidence Interval	Result	95% Confidence Interval
Sensitivity	0.80	(0.38, 0.96)	0.60	(0.23, 0.88)
Specificity	0.78	(0.45, 0.94)	0.50	(0.19, 0.81)
Positive Predictive Value (PPV)	0.67	(0.30, 0.90)	0.50	(0.19, 0.81)
Negative Predictive Value (NPV)	0.88	(0.53, 0.98)	0.60	(0.23, 0.88)
Positive likelihood ratio (LR+)	3.60	(0.98, 13.19)	1.20	(0.41, 3.51)
Negative likelihood ratio (LR–)	0.26	(0.04, 1.54)	0.80	(0.21, 3.05)

Table 7.

Wound bed fluorescence and histological status of margins in Phase 2.

Tumor type	Pattern of residual fluorescence	Number of adjunctive resected tissue	Histological status of adjunctive resected tissue	Histological status of margins of resected tumor
STS	Single spot	2	Neoplastic	Infiltrated
STS	Single spot	1	Non-neoplastic	Non-infiltrated
STS	Single spot	3	Neoplastic	Infiltrated
STS	Single spot	3	Non-neoplastic	Non-infiltrated
STS	Single spot	2	Neoplastic	Infiltrated
STS	Single spot	3	Neoplastic	Infiltrated
MCT	Single spot	1	Non-neoplastic	Non-infiltrated
MCT	Single spot	1	Non-neoplastic	Infiltrated
MCT	Single spot	0	Not performed	Infiltrated
MCT	Diffuse	0	Not performed	Infiltrated
MCT	Diffuse	0	Not performed	Non-infiltrated
MCT	Diffuse	0	Not performed	Non-infiltrated

Considering the 8 wound beds in which adjunctive fluorescent tissue was resected (6 STSs and 2 MCTs), their histopathological evaluation found neoplastic tissue only in 4 STSs all of which had also infiltrated margins in the excised tumor. (Table 7).

The LR⁺ of the STS showed that the probability of having a fluorescent wound bed for infiltrative margins was 3.60 times the probability to have the same result for non-infiltrated margins; for MCT the LR⁺ value was only 1.20, suggesting a best performance for STS.

The LR⁻ of the STS showed that the probability of non-fluorescent tumor bed for non-infiltrated margins was 3.85 times the probability to have the same result for infiltrate margins; for MCT the LR⁻ value was only 1.25, suggesting a best performance for STS.

Nevertheless, on the basis of the classification criteria reported in the method section, these results indicated that using only the fluorescence gave a weak indication about the “real” status of margins, but the 95% confidence interval is wide for the LR⁺ (0.98, 13.19) and LR⁻ (0.04, 1.54); considering the superior value of the confidence intervals there is an indication of a possible excellent performance of the test (LR⁺>10 and LR⁻<0.1) in STS. For MCT, the 95% confidence interval may suggest a good test performance (LR⁺: 0.41, 3.51; LR⁻ 0.21, 3.05).

The trend of PPV for STS and MCT in function of the prevalence for different scenario of STS and MCT shows that for a higher value of sensitivity and specificity the PPV was higher for all the values of prevalence. The highest PPV for the same combination of sensitivity and specificity is for a higher prevalence (Fig. 1).

Fig. 1.

Positive Predictive Value in function to prevalence of infiltrated margins in STS and MCT. Panel (a) shows the results for STS and panel (b) shows the results for MCT. Sens=Sensitivity. Spec=specificity.

The trend of NPV for STS and MCT in function of the prevalence for different scenario of STS and MCT shows that for a higher value of sensitivity and specificity the NPV was higher for all the values of prevalence. The highest NPV for the same combination of sensitivity and specificity is for the lower prevalence (Fig. 2).

Fig. 2.

Negative Predictive Value in function to prevalence of infiltrated margins in STS and MCT. Panel (a) shows the results for STS and panel (b) shows the results for MCT. Sens=Sensitivity. Spec=specificity.

The best cut-off according to Youden index for the STS was 0.548 (Sensitivity = 0.78 and Specificity = 0.77, LR⁺=3.39, LR⁻=0.29). The area under the ROC curve for STS was 0.809. Therefore, the SBR in the wound bed has a good discriminative ability for the histological margins for this type of tumor. The area under the ROC curve for MCT was 0.506. Therefore, the SBR in the wound bed has a very poor discriminative ability for the histological status of margins for this type of tumor. Given the extremely poor discrimination, the best cut off was not considered (Figs. 1, 2 and 3).

Fig. 3.

ROC curve for infiltration margin identification by fluorescence in the wound bed in STS and MCT. Panel (a): STS. Panel (b): MCT. TPR=True Positive Rate. FPR=False Positive Rate. The gray point in panel ‘’a’’ identifies the position of the Youden index.

Discussion

Considering the growing interest and application of NIRF-ICG in small animal surgical oncology^21–24, this study aimed to investigate the clinical use of free-ICG in canine surgical oncology, particularly for STS and MCT, to increase the likelihood of complete tumor removal^2,3. Targeted fluorophores are not currently in use in veterinary medicine clinical setting and few studies have investigated the use of free-ICG for TFGS^18,25,26; however, early research differ from our work as they employed higher doses of ICG and focused either on a single tumor histotype or on multiple histotypes with limited sample sizes^25–27. None of these studies explored the performance of the technique across different spontaneous cancers. Given the recent advancements in commercially available NIRF technologies, updated information on standardized TFGS using free-ICG are warranted. The findings of this pilot study indicate that the use of free-ICG, combined with intraoperative real-time quantification of fluorescence at the applied dose and timing and TBR or SBR cut-off chosen, provide a substantial support to the surgeon during STS and MCT removal. Nonetheless, differences observed between these histotypes highlight the need for further assessments and more detailed discussion.

Regarding feasibility, the results of this study demonstrate that using a dose of 0.5 mg/kg of ICG administered intravenously 24 h before surgery makes TFGS practicable for both STS and MCT. Feasibility was high for both histotypes, with 100% for STS and 79% for MCT. Literature shows a significant heterogeneity with regard to the ICG dose (0.5–7.5 mg/kg), timing of administration (1–24 h preoperatively), TBR and SBR cut off (1.2–3.2) and fluorescence quantification devices, in both dogs and humans^{13,14,18,20,25,26,35,36}. This variability poses a major challenge for standardizing and reproducing TFGS protocols. In small animal surgical oncology, the limited available data further complicate the establishment of consensus on optimal ICG parameters³⁶. Previous studies in dogs have employed higher ICG doses (3–7.5 mg/kg)^{25–27,35,37}, whereas our study adopted a lower dose (0.5 mg/kg) that has been already validated in humans²⁰, possibly reducing dose per kilogram administered and consequently the costs related to the use of this fluorophore in daily clinical practice. The 24-hour timing from administration to surgery is frequently reported in both human and canine literature^13,14,26,37; however, the optimal window should be defined based on tumor histotype in future studies. Despite the overall good feasibility, the technique’s performance differed between the two tumor types: the TFGS proved consistently successful in all included STS cases, whereas a 21% (3/14) failure rate was observed in MCTs. It cannot be excluded that these neoplasms may require different administration protocols. Indeed, while the selected dose and timing proved consistently optimal for STS, MCTs may benefit from different parameters to allow the TGFS in all tumors. Moreover, the recorded TBR was higher in the STSs group compared to MCTs, although no statistically significant difference was observed between the two distributions, possibly due to the limited sample size. This suggests that STSs are more easily distinguishable from surrounding tissues than MCTs at this dosage, timing and TBR, possibly due to differences in tumor biology affecting the EPR effect³⁸. Increased vascular permeability in tumors allows ICG to extravasate and accumulate in the tumor microenvironment. However, in cases with marked peritumoral inflammation, common in MCTs due to mast cells degranulation and histamine release, ICG may accumulate extensively in peritumoral tissues, thus extending beyond the gross tumor margin and limiting the effectiveness of non-targeted NIRF agents like ICG³⁷.

Difference between STS and MCT were also observed in the assessment of wound bed residual fluorescence (SBR in Phase 2) and related to the histopathological status of margins. In the STS population, the technique demonstrated good precision in distinguishing among infiltrated and non-infiltrated margins. In detail, NIRF-ICG in STS cases showed good sensitivity and specificity, at 80% and 78%, respectively, whereas in MCTs it had lower sensitivity and specificity values (60% and 50%, respectively) (Table 6; Figs. 1, 2 and 3). Nevertheless, the estimated performance of the test evaluating by LR values was based on a very limited case series and should be revaluated to obtain a higher precision of the estimate. These findings suggest that, in MCTs, the ability of NIRF-ICG to accurately differentiate between infiltrated and non-infiltrated margins seems to be limited. This may be due to the same reason related to biological behavior of MCT that affected the results of TBR³⁷. Moreover, ROC curve analysis further confirmed this discrepancy among tumor types. In STSs, the AUC value of 0.80 indicates good discriminative ability, meaning that infiltrated margins generally exhibited higher SBR values than non-infiltrated ones, making NIRF-ICG a reliable intraoperative tool in STS cases. On the other hand, the AUC for MCTs was close to 0.50, indicating that the technique performed no better than chance in detecting residual neoplastic tissue in the wound bed. (Fig. 3).

An adjunctive tissue resection based on intraoperative fluorescence signals detected before wound bed reconstruction was performed in a total of 6 STS and 2 MCT cases (Tables 3 and 7). Histological analysis confirmed the presence of cancer in the fluorescent adjunctive tissue removed in 4 STSs and all of these cases had infiltrated margins. Since to date the result of histological status of the margins is not available immediately after tumor excision, the use of real time intraoperative quantitative assessment of fluorescence may help the surgeons to remove adjunctive neoplastic tissue otherwise not identifiable without this technology, contributing to reduce the burden of microscopic disease^13,15,32.

Among the objectives of this pilot study, we also explored the influence of TFGS with NIRF-ICG in intraoperative decision-making. The results showed that in 64% of STS and 55% of MCT an increased tissue resection was performed in real-time based on the detection of intensity of fluorescence during excision of tumors (Phase 1) and evaluation of the wound bed (Phase 2) (Table 5). These findings, together with the histopathological results and the higher sensitivity and specificity, suggest that TFGS may have greater clinical utility in STS compared to MCT at this dosage, timing and TBR. The good performance of TFGS in STS confirms what has been reported in the literature on human sarcomas, where the use of NIRF-ICG appears to be a useful technique to support surgical decision-making and improves the ability of obtaining non infiltrated margins in soft tissue and bone sarcoma surgery^13–15,20.

The influence of TFGS with NIRF-ICG on intraoperative decision-making should also be considered in light of the technology adopted in this study, which operates without the need for additional personnel beyond the surgical team and does not require external software or separate devices, as all functionalities are fully integrated into the intraoperative system. This integration enabled a quantitative evaluation of fluorescence intensity, through the calculation of TBR in Phase 1 and SBR in Phase 2, thus ensuring more accurate real-time intraoperative decision-making by the surgeons. Nevertheless, our results indicate that the TBR and SBR value seems to be appropriate and consistent with previously published results, in particular for STS^20,33. The Youden index identified a cut-off value of 0.54 as the best balance between sensitivity and specificity (Fig. 3). However, this threshold did not appear to offer superior diagnostic performance compared to the clinically established cut-off of 2. Given the small sample size, these findings should be interpreted with caution. Further studies involving larger cohorts are needed to validate the optimal diagnostic threshold and support its clinical applicability.

Despite the relatively small sample size, the number of cases included is comparable to that reported in previous studies in dogs^11,25–28. This limitation inherently constrains the generalizability of our findings and underscores the need for confirmation in larger caseloads. Such studies should also explore potential differences in fluorescence performance across STS subtypes or grades, as well as between cutaneous and subcutaneous MCTs, and better quantify the extent of the increased margins. Furthermore, to improve standardization of dose, timing, and TBR for each histotype, dedicated pharmacokinetic studies of ICG are also warranted.

In conclusion, this study confirms the clinical applicability of TFGS using NIRF with free-ICG for the excision of canine STS and MCT. However, the technique demonstrated superior performance in STS, likely due to biological differences influencing ICG accumulation. In this context, the integration of intraoperative real-time quantification of NIRF-ICG into surgical decision-making may represent an important first step toward future investigations aimed at optimizing free-ICG use for TFGS in canine surgical oncology and enhancing its use in daily clinical practice.

Materials and methods

Study population

This observational pilot study was approved by the Institutional Ethical Committee for Animal Care of the University of Milan (OPBA 02–2025). This study enrolled client-owned dogs referred to the Veterinary Teaching Hospital of the University of Milan from February to August 2025 bearing spontaneous MCT (cutaneous and subcutaneous) and subcutaneous STS at first presentation or recurrence. Owners signed a written informed consent for ICG administration, surgical procedures, and data collection. All procedures were carried out in accordance with the Italian National legislation for animal welfare (DL 14th March 2014 n.26) and the best standards of veterinary clinical practice, as defined by the Good Veterinary Practice guidelines of the Federation of Veterinarians of Europe and the Italian National Federation of Veterinary Orders (FNOVI, 29th January 2005) and in accordance with ARRIVE guidelines.

Dogs were included in the TFGS with NIRF-ICG if presenting with a macroscopical tumor mass, with a preoperative cytological or histological diagnosis of STS or MCT, eligible for curative intent surgery. Dogs presenting with simultaneous multiple tumors were included, as long as tumors were located in distinct anatomical regions and on opposite sides of the body, thereby not involving the same surgical field. Exclusion criteria were: the presence of distant metastases and/or scars from previous tumor excision with residual microscopic disease. Tumors in which margin assessment with TFGS was deemed clinically irrelevant (i.e., limb or tail amputation for distally located tumors, where resection had to be performed at a considerable distance from the lesion) were also excluded.

Complete blood count with differential and serum biochemistry were always performed to assess general anesthesia risk and eligibility for the intravenous ICG administration protocol. Dogs with American Society Anesthesiologist physical status superior than 3 were excluded to avoid potential side effects associated with ICG administration³⁹.

ICG administration and surgical planning

The solution of ICG (Verdye, DiagnosticGreen GmbH, Aschheim, Germany) was prepared with 10 ml of sterile water for injection, obtaining a concentration of 2.5 mg/ml. A dose of 0.5 mg/kg of ICG was administered intravenously 24 h before the planned surgery. Occurrence of any adverse events during and after ICG intravenous administration was recorded and categorized in accordance with LeBlanc and coll⁴⁰. In all cases, sentinel lymph node mapping was performed prior to tumor excision using lymphoscintigraphy; peritumoral injections of methylene blue were avoided in sentinel lymph node mapping to prevent interference of the dye with tumor margin assessment.

Surgical dose adopted was classified as: radical (excision of an entire anatomical compartment); wide excision (2–3 cm lateral for STS and MCT > 2 cm or proportional lateral margins for MCT < 2 cm only, and at least 1 involved fascial layer for deep margins including bone structures)^7,8; planned narrow excision (if wide excision was not possible, the widest margins were applied based on tumor dimension and localization)^6,9,10,41,42. On the day of surgery, wide clipping of the surgical area was performed, and the surgical field was prepared with sterile chlorhexidine-based scrubbing. Before NIRF-ICG application, the surgeons outlined the planned lateral skin excision with a sterile skin marker, as usually performed, considering the surgical dose, palpation findings of deeper tissue involvement, and, if available, pre-surgical computed tomography imaging. Electrocautery was deliberately avoided during tissue resection to prevent thermal artifact formation and deposition of charred tissue remnants, which might compromise the accuracy of fluorescence imaging and subsequent analysis of the surgical bed⁴³. All surgical procedures were performed by the same surgeon (DS) with several years of experience in clinical research of canine STS and MCT and in the use of NIRF technology. Occurrence of intra-operative or post-operative complications was recorded and classified according to Le Blanc and coll⁴⁰.

NIRF technical information

The NIRF device used was SPY-PHI QP% system (Stryker; MIDA, Tecnologia Medica S.p.a.) with a Portable Handheld Imager (SPY-PHI). The SPY-PHI has a dedicated sterile plastic drape to be intraoperatively handheld by the surgeons.

In addition to the white light visualization mode, the SPY-PHI QP% system (Stryker) allows three visualization modalities for fluorescence, defined as overlay mode, color maps mode, and contrast mode (Fig. 4). These modalities can be freely selected and switched interchangeably during surgery, directly from the probe. The quantification of the fluorescence intensity of the regions of interest (ROI) in real-time is allowed by the QP% software (Stryker).

Fig. 4.

Visualization modalities available for the SPY-PHI QP% system (Stryker). (A) The white light visualization - SPY-PHI QP% system is used as a normal camera; (B) the overlay mode – the fluorescence is represented with the overlaying green color. Increasing intensity of green color reflects increasing recorded fluorescence from the probe; (C) the color map mode – intensity of fluorescence is reported on a chromatic scale ranging from red (maximum fluorescence intensity) to blue (minimum fluorescence intensity). Grey color indicates absence of fluorescence.; (D) the contrast mode – intensity of fluorescence is represented on a monochromatic scale, higher fluorescence intensity is white, while the lower is black. All reported fluorescence intensities vary based on the set baseline.

For this study, immediately before skin incision, the handheld camera was positioned 10–40 cm above the tissue of interest, and surgical lights were turned off when using the SPY-PHI device. At the beginning of each evaluation, the white balance was done, then the QP% software was calibrated using the tumor tissue as the reference baseline. All subsequent fluorescence measurements were quantified in relation to the baseline signal, ensuring case-by-case consistent and tumor-centered assessment throughout the procedure (Fig. 5).

Fig. 5.

Tumor-to-background ratio (TBR) and signal-to-background ratio (SBR) assessment in the two phases of TFGS. In this picture is reported the schematic representation of the real time calculation of TBR and SBR during the two phases of TFGS. All the clinical pictures reported are in color map modality. A-B: The QP system is used to assess the TBR for the TFGS feasibility. (A) The QP system is set on the tumor (100%); (B) the relative background quantification is assessed (5%). The TBR is calculated as the ratio between the tumor and the background fluorescence quantification. In this particular case, TBR value resulted > 2 (100%/5%= 20) therefore the technique was defined as successful. C-D: the QP system is used to assess the SBR of the tissue to be dissected (ROI). During the excision the fluorescence intensity of the ROI (the site of dissection) (C) and the background (normal tissue) (D) is real-time assessed and the SBR calculated in order to perform the dissection in the SBR ≤ 2. In this particular case the SBR is calculated as 6%/8%= 0.75) E-F: the QP system is used to assess the SBR of the wound bed after the tumor removal. The wound bed fluorescence intensity (representing the ROI in this case) (E) and the background (F) are real time assessed and the SBR calculated. In this case the SBR is ≤ 2 (SBR = 7%/6%= 1,16) and the wound bed has no residual fluorescence.

Using the QP% software, the fluorescence intensity of the tumor, each region of interest (ROI), and the background were recorded in real time. The ‘background’ was defined before excision as grossly normal-appearing tissue adjacent to the tumor, while the ‘signal’ was defined as the fluorescence intensity of a specific ROI. The tumor-to-background ratio (TBR) was calculated as the ratio of the tumor fluorescence intensity to that of the background and used to assess the feasibility of the procedure (see below). The signal-to-background ratio (SBR) was calculated as the ratio of the signal from a ROI to the background. The SBR was used for the decision-making of the dissection of skin and fascial plane, and to assess residual fluorescence in the wound bed. A TBR and SBR cut-off value of 2 was arbitrarily chosen, as it is, to the authors’ knowledge, the most frequently reported quantification threshold in both human and veterinary literature for distinguishing neoplastic from normal tissue^19,27,33. (Fig. 5).

TFGS procedure

The TFGS procedure consisted of the following two phases:

Phase 1: tumorsurgical resection

The tumor fluorescence was real-time quantitatively checked with the SPY-PHI. Using the QP% software, in order to assess the feasibility of the technique, the TBR was calculated. If the TBR was > 2 (the tumor was considered distinguishable from the surrounding tissue and TFGS feasible). If TBR was ≤ 2 before and after skin incision, the TFGS was considered as failed, the surgery was performed as planned traditionally⁴⁴, and, therefore, the case was excluded from the following TFGS procedures. The SBR was used throughout the entire excision phase to guide the surgeons in tumor removal, thanks to the real-time quantification system, which was always intraoperatively available. Dissection around the tumor was performed, whenever possible, in tissues with SBR ≤ 2, as they displayed low fluorescence intensity and most likely non-neoplastic tissue. During all the excision phase, the real-time quantification system for TBR and SBR calculation was always available to guide the surgeons.

Furthermore, it was recorded whether TFGS led to the removal of small additional amounts of tissue in the lateral skin and/or subcutaneous fascia (≥ 0.5 cm) and/or deeper tissues (additional fascial plane), with respect to the area originally planned.

Phase 2: wound bed fluorescence assessment before wound reconstruction

After the tumor removal, maintaining the original setting for QP% (100% on tumor fluorescence), the fluorescence quantification of the wound bed was checked with the SPY-PHI. Using the QP% software the SBR was calculated: the signal of the wound bed was compared to the background and an SBR > 2 was interpreted as residual fluorescence in the wound bed; an SBR ≤ 2 was interpreted as absence of residual fluorescence in the wound bed²⁵(Fig. 6). When residual fluorescence was present, it was classified as diffuse (if extensively diffuse on the tumor bed) or single spot (if only a single spot of fluorescence was visible). Wound bed residual fluorescence (spot or diffuse) was removed (adjunctive resection and/or incisional biopsy), if feasible and without increasing the risk of morbidity. All adjunctive tissues removed were individually submitted for histopathological evaluation.

Fig. 6.

Examples of wound bed analysis for residual fluorescence. (A) the wound bed has no residual fluorescence (SBR ≤ 2); (B) the wound bed is positive for residual fluorescence (SBR > 2).

The surgical aid of TFGS was evaluated by asking the surgeon whether NIRF imaging assisted intraoperative decision-making, leading to an increase in the planned resection (phases 1–2)¹³.

After phases 1–2, the surgical procedure continued with wound reconstruction.

Histopathological analysis

All specimens were promptly oriented by the surgeon post-excision and submitted for histopathological evaluation. For soft tissue sarcomas (STS), tumor grade was assessed^45,46. For mast cell tumors (MCT), cutaneous forms were graded using both the Patnaik and Kiupel systems^47–49, while subcutaneous MCTs were classified according to Thompson et al. (2011)⁵⁰. A complete longitudinal section was examined, and the status of surgical margins was assessed using 3-dimensional technique^51,52. Histological margins were categorized as infiltrated and non-infiltrated.

Data collection and statistical analysis

Recorded data about dogs signalment were the following: breed, sex, age, body weight, body condition score (BCS) on a five scale (classified in underweight if 1–2/5; normal weight if 3/5; overweight if 4–5/5). Clinical variables of tumors included: presentation (first or recurrence), anatomical location (categorized as limbs - below the elbow/stifle, excluding digits; trunk - upper the elbow/stifle, included thoracic mammary glands; head and neck; inguinal - perineal, genitals, and inguinal mammary glands; digits; and tail), dimension (in millimeters), presence of tumor ulceration, pigmentation of the overlying skin, clinical skin mobility overlying the tumor and if tumor was fixed or not fixed to deeper fascia. Surgical data included: surgical dose (radical, wide excision or planned marginal excision); tumor surgical timing (calculated in minutes from skin incision to last skin suture); occurrence of intra or post-operative complication and grade⁴⁰.

Tumor’s histopathological variables collected were: histotype, grade (when applicable), status of margins (infiltrated vs. non-infiltrated).

Based on the TFGS, the following data were collected: occurrence of adverse events after ICG intravenous administration; the TBR of the tumor; the feasibility of the technique (defined as TBR > 2); whether the fluorescence confirmed or increased the resection (either the excision of ≥0,5 cm of one or more lateral margins and/or deeper fascial plane, or via additional resection of adjunctive fluorescent tissue in the wound bed); the SBR of the wound bed post-excision to confirm presence/absence of residual fluorescence.

The distribution of retrieved variables was synthesized by a descriptive statistic: for categorical variables, the frequency of each modality was reported as the % of the total cases, for continuous variables, was expressed as median (first quartile; third quartile) since the distribution were asymmetrical. The cumulative distributions of TBR for STS and MCT were compared using the exact two-sample Kolmogorov-Smirnov test. The null hypothesis of this test was that the two groups were sampled from the same population distribution. The same test was also used to compare the SBR of STS and MCT. Exact distributions for the two-sample (Smirnov) test were computed by the algorithm proposed by ⁠Schröer and Trenkler (1995)⁵³.

The performance of the TFGS to identify the status of the margins, was evaluated considering the histopathological marginal status of the reference standard (gold standard). The performance measures for each tumor category (STS and MCT) were sensitivity (i.e. the probability of residual fluorescence given infiltrated histopathological margins) and specificity (i.e., the probability of absence of residual fluorescence given non infiltrated histopathological margins). For sensitivity and specificity, the confidence intervals were calculated according to Wilson⁵⁴. This approach is recommended in cases of small sample size (n < 40) due to its superior coverage probability, especially when the proportion values (such as sensitivity or specificity) are not extremely close to 0 or 1⁵⁵. Based on sensitivity and specificity, positive and negative likelihood ratios (LR⁺, LR^-) were considered. LR + is the ratio between sensitivity and 1- specificity (i.e. the ratio between the probability of the presence of residual fluorescence given infiltrated histopathological margins and the probability of residual fluorescence given the non-infiltrated histopathological margins). The higher the LR⁺ value, the more effective fluorescence is in identifying positive margins. LR- is the ratio between 1-sensitivity and specificity (i.e. the ratio between the probability of the presence of residual fluorescence given non infiltrated histopathological margins and the probability of no residual fluorescence given the non-infiltrated histopathological margins). The lower the LR^- value, the more effective fluorescence is in ruling out infiltrated margins. The strength of the effectiveness of fluorescence in identifying the histopathological status of the margins can be quantified according to the criteria suggested by Ebell⁵⁶: LR⁺ ≥ 10 relevant (and often conclusive) result regarding the presence of infiltrated margins; 5 ≤ LR⁺ < 10 → moderate indication of the presence of infiltrated margins; 2 ≤ LR⁺ < 5 → weak indication of the presence of infiltrated margins. LR^- ≤ 0.1 relevant (and often conclusive) result regarding the presence of non-infiltrated margins 0.1 < LR^- ≤ 0.2 moderate indication of the presence of non-infiltrated margins 0.2 < LR^-≤ 0.5 weak indication of the non- infiltrated margins. For the likelihood ratios, confidence intervals were based on the formulae provided by Simel at al⁵⁷. All those computations were performed using the epi.tests function from the R package epiR, version 2.0.76⁵⁸. Based on reported infiltrated margins prevalence ranges for STS (16–69.5%)^{1,9,10,59–61} and MCT (15–56%)^{1,10,50,62–65} the positive predictive value (PPV) and negative predictive value (NPV) were calculated under three scenarios: using the point estimates of sensitivity and specificity, using the lower limits of their 95% CIs (representing a worst-case scenario), and using the upper limits of the CIs (best-case scenario). This was done to explore the impact of diagnostic accuracy variation across different prevalence levels. The confidence intervals for the predictive values were calculated according to Wilson⁵⁴.

Lastly, the diagnostic performance of residual fluorescence at the wound bed (SBR) without considering the cut-off value was evaluated using receiver operating characteristic (ROC) curve analysis. Given the small sample size, a non-parametric method based on smoothed empirical distribution functions was employed, as recommended by Jokiel-Rokita and Pulit (2013)⁶⁶. To identify the best cut-off, the Youden index was considered. The Area under the ROC evaluated the discriminant ability of fluorescence. All the analysis were performed with R software version 4.4.0⁶⁷.

All procedures performed during the TFGS with NIRF-ICG and combination with histologic results are summarized in Fig. 7.

Fig. 7.

Decision making process during TFGS. TBR= tumor-to-background ratio, SBR= signal-to-background ratio, NIRF-ICG=near infrared fluorescence with indocyanine green, FP=false positive, TP=true positive, TN=true negative, FN=false negative.

Author contributions

Conceptualization, E.M.G.; D.S.; A.U; Investigation: E.M.G.; D.S.; A.U.; R.F.; L.A, P.R.; S.D.; C.G; V.G; C.R; Data curation: E.M.G.; P.B; E.L; Methodology: P.B; E.L.; Writing – Initial Draft: E.M.G.; D.S.; A.U. Writing-Review and editing: E.M.G.; D.S.; A.U.; R.F.; L.A, P.R.; S.D.; C.G; V.G; C.R; F.F; F.B. All authors reviewed the manuscript.

Data availability

The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.