Access the full text.
Sign up today, get DeepDyve free for 14 days.
www.nature.com/npjbcancer ARTICLE OPEN Intratumoral in vivo staging of breast cancer by multi-tracer PET and advanced analysis 1 1,2 3 1 1 1 4 Jennifer Griessinger , Julian Schwab , Qian Chen , Anna Kühn , Jonathan Cotton , Gregory Bowden , Heike Preibsch , 1,5 5,6 3 3 6 3,7 Gerald Reischl , Leticia Quintanilla-Martinez , Hidetoshi Mori , An Nguyen Dang , Ursula Kohlhofer , Olulanu H. Aina , 3 1,5,8 3 1,5 Alexander D. Borowsky , Bernd J. Pichler , Robert D. Cardiff and Andreas M. Schmid The staging and local management of breast cancer involves the evaluation of the extent and completeness of excision of both the invasive carcinoma component and also the intraductal component or ductal carcinoma in situ. When both invasive ductal carcinoma and coincident ductal carcinoma in situ are present, assessment of the extent and localization of both components is required for optimal therapeutic planning. We have used a mouse model of breast cancer to evaluate the feasibility of applying molecular imaging to assess the local status of cancers in vivo. Multi-tracer positron emission tomography (PET) and magnetic resonance imaging (MRI) characterize the transition from premalignancy to invasive carcinoma. PET tracers for glucose consumption, membrane synthesis, and neoangiogenesis in combination with a Gaussian mixture model-based analysis reveal image-derived thresholds to separate the different stages within the whole-lesion. Autoradiography, histology, and quantitative image analysis of immunohistochemistry further corroborate our in vivo ﬁndings. Finally, clinical data further support our conclusions and demonstrate translational potential. In summary, this preclinical model provides a platform for characterizing multistep tumor progression and provides proof of concept that supports the utilization of advanced protocols for PET/MRI in clinical breast cancer imaging. npj Breast Cancer (2022) 8:41 ; https://doi.org/10.1038/s41523-022-00398-x INTRODUCTION molecular imaging modalities have the potential to noninvasively provide spatially resolved functional whole-lesion information. Accurate identiﬁcation of high- and low-risk neoplasms and Positron emission tomography (PET) is established in daily practice monitoring their progression is currently one of the more in the ﬁelds of lung cancer diagnostics, prostate cancer challenging problems in breast cancer. Risk is assessed primarily 5–7 diagnostics, or lymphoma . Promising data have also been by tumor stage (size and spread including metastases) and tumor produced in breast cancer using molecular imaging to target grade (in breast, a standardized histologic score encompassing 8–14 glucose metabolism, proliferation, or receptor status .A proliferation and morphology). Tumor grade and phenotypic complete molecular characterization of tumor heterogeneity heterogeneity as well as coincident invasive and in situ lesions within single lesions can be important for clinical decision-making; require detailed study of the pathology for accurate reporting but however, PET measurements have not been standardized in this remains subjective. Our current classiﬁcation schema fails to breast cancer. consider the indolent phenotype, leading to overdiagnosis and Many invasive ductal carcinomas (IDCs) also referred to as overtreatment . Swanton described the phenomenon of intratu- invasive mammary carcinomas of no special type (NST) are moral heterogeneity as a “process through time and space” in associated with components of ductal carcinoma in situ (DCIS). which alterations and mutations occur on different time scales in DCIS without an invasive component can sometimes progress various regions of the same tumor . Accurate comprehensive localized staging of breast lesions including the size and over time to invasive carcinoma, but the time interval can be distribution of in situ and invasive areas of different grade/ decades and in many cases progression may stall indeﬁnitely. phenotype could be an important step in clinical management. Some already consider the presence of a DCIS component within 15–17 an invasive breast cancer a positive prognostic marker . The staging process is well deﬁned and crucial for subsequent Therefore, differentiating between pure NST, DCIS, and mixed treatment decisions . The current standard procedures for staging breast cancer include clinical examination, imaging by mammo- forms of NST/DCIS with noninvasive methods may be an graphy and/or sonography, sometimes accompanied by magnetic important factor in clinical and preclinical evaluations. Like all resonance imaging (MRI), and histopathological analysis of biopsy clinical studies, validations of these results are necessarily and excision tissue specimens. Pathologists, limited to the local dependent on the evaluation of large cohorts and statistical “snapshot” nature of a biopsy specimen, are well aware of tumor analysis. heterogeneity. They report the highest grade found after In order to demonstrate the power of in vivo molecular imaging to characterize intralesional heterogeneity, we employed multi- examining multiple tissue sections from any tumor biopsy . Beyond biopsy-based staging and morphological imaging, parametric PET/MRI to the transgenic polyomavirus middle T (FVB/ 1 2 Werner Siemens Imaging Center, Department of Preclinical Imaging and Radiopharmacy, Eberhard Karls University Tuebingen, Tuebingen, Germany. Institute of Medical 3 4 Systems Biology, Ulm University, Ulm, Germany. Center for Immunology and Infectious Diseases, University of California, Davis, CA, USA. Department of Radiology, University Hospital Tuebingen, Tuebingen, Germany. Cluster of Excellence iFIT(EXC 2180) “Image-Guided and Functionally Instructed Tumor Therapies”, University of 6 7 Tuebingen, Tuebingen, Germany. Department of Pathology, Eberhard Karls University Tuebingen, Tuebingen, Germany. Janssen Pharmaceutical, Spring House, PA, USA. German Cancer Consortium (DKTK), Partner Site Tuebingen; German Cancer Research Center (DKFZ), Heidelberg, Germany. email: A.Schmid@med.uni-tuebingen.de Published in partnership with the Breast Cancer Research Foundation 1234567890():,; J. Griessinger et al. N-Tg(MMTV-PyVT)634Mul/J) derived mammary intraepithelial heterogeneous uptake within the individual lesions (Suppl. Figure neoplastic outgrowth (MIN-O) model . This allowed the in vivo S2a, Fig. 1). evaluation of heterogeneous cell populations during neoplastic Advanced, voxelwise analysis of summed [ F]FDG uptake in all progression. lesions at each time point demonstrated a shift in voxel values The transplanted MIN-O grows contact-inhibited within the over time towards higher uptake (Fig. 1a). As the summed mouse mammary fat pad . On histological examination, the MIN- histogram of all time points covered all uptake populations that Os do not form normal branching mammary trees, but form appeared over the entire study duration, this summed dataset was ﬁtted using a Gaussian Mixture Modell (GMM). The ﬁtting criteria abnormal hyperplastic outgrowths . The most peripheral growing Akaike Information Criterion (AIC) and Bayesian Information edge forms modiﬁed terminal end buds that extend into the fat Criterion (BIC) deﬁned a sum of ﬁve Gaussian distributions as pad but lack the orderly organization of the normal terminal end the best ﬁt for the study (Fig. 1b, c). Applying the corresponding bud . The layer behind the growing edge contains more thresholds to the single time points demonstrated the appearance differentiated cells which typically form disorganized ducts with of the populations over time (Fig. 1d). The ﬁrst population with irregular alveoli. As the transplant grows, an inner core of ducts uptake values < 1.7%ID/cc was identiﬁed as a peripheral and alveolar structures extend. This zone is heterogenous and, background population. The following two populations with broadly described, has more well-differentiated areas of glands values of 1.7–3.6%ID/cc and 3.6–6.7%ID/cc could not be separated with abundant eosinophilic cytoplasm and areas of differentiated by in vivo imaging, as both populations already existed at the hyperchromatic dysplastic alveolar cells. These hyperchromatic early time point w4 when only fat tissue and hyperplasia were dysplastic cells are, CA-IX positive and could be considered “high present within the fat pads. Whether the fat corresponds to one nuclear grade” MIN lesions. Previous work employing the same population and the hyperplasia to another could not be veriﬁed MIN-O mouse model described the peripheral growing edge as due to the spatial resolution of the PET (1.6 mm maximal “proliferation zone” and the encapsulated central region as achievable resolution ). Therefore, these two populations were “differentiation zone”, where the invasive adenocarcinoma summed together as the fat and hyperplasia population in blue. develop . Thus, alike the human disease, this model undergoes The fourth population in green (6.7–11.8%ID/cc) appeared clearly a histologically identiﬁable, multistep neoplastic progression to in w8 when histological analysis identiﬁed DCIS-like MIN lesions 18,20–24 DCIS-like, premalignant MIN and invasive carcinoma (IC) (Suppl. Figure S2b). The remaining population in red with uptake (Suppl. Figure S1). In this study, PET biomarkers have been used to values > 11.8%ID/cc emerged in a high amount at late time points detect changes in tumor metabolism over time in the same in tumor development when the tumors showed invasive growth animals to track the transition to malignancy. Our panel of in a 3-dimensional manner (Fig. 1d & e, Suppl. Figure S2b). The 18 18 biomarkers included [ F]ﬂuoro-2-deoxy-2-D-glucose ([ F]FDG), presence of high-uptake populations (red and green) at earlier 11 11 [ C]choline ([ C]Chol), and quantiﬁed the expression of α β - V 3 time points could be explained by single lesions that presented 68 68 25 integrin with [ Ga]Ga-NODAGA-c(RGDfK) ([ Ga]RGD) . For each faster development and early invasiveness (mouse M1, Fig. 1e). time point, autoradiography with subsequent staining of the slides Autoradiography results conﬁrmed increasing uptake of [ F]FDG, co-registered PET tracer uptake with histopathology. Advanced from low (w4) and moderate uptake in premalignant regions analytic techniques for identifying intratumoral heterogeneity (especially in w8) to the highest uptake in invasive carcinoma were applied to the data and provided unique insight into the (especially in w11) (Fig. 1f). dynamics of neoplastic progression. Our studies also included pre- lactating and lactating mammary gland controls to document Metabolic characterization of tumorigenesis normal physiological processes of increased metabolism and Mean value analysis. Following the intratumoral staging proliferation. Finally, exemplary patient data from a study on 18 approach using [ F]FDG, subsequent investigation addressed intratumoral heterogeneity in [ F]FDG PET/MRI of primary breast proliferation and angiogenesis during tumor development. Five cancer patients were retrospectively analyzed to assess and mice were measured at the time points w3, w7, w10, and w13 with illustrate the results in a clinical setting. 18 11 68 MRI, [ F]FDG, [ C]Chol, and [ Ga]RGD. Five additional mice were only measured with [ F]FDG for subsequent ex vivo analysis at each time point. Prelactating and lactating glands served as RESULTS control tissue (Fig. 2). Intratumoral staging using [ F]FDG-PET H&E histology demonstrated tumor development in compar- To differentiate neoplastic stages during tumor development from ison to healthy, prelactating, and lactating mammary glands (Fig. premalignant DCIS-like MIN to IC, we investigated the glucose 2a). The pre- and lactating glands, as well as the early lesion stages 18 11 metabolism of lesions at different time points during tumorigen- in w3, presented no or very low uptake of [ F]FDG, [ C]Chol, and 18 68 esis by using [ F]FDG PET (Suppl. Figure S2a). Four weeks after [ Ga]RGD. In contrast, the latest stage (w13) exhibited high MIN-O transplantation (w4), hematoxylin and eosin (H&E) histol- uptake with a clear heterogeneous pattern (Fig. 2b). A mean value analysis of prelactating glands showed tracer ogy and whole-mount staining showed that the fat pads were uptake in the range of background accumulation (Table 1). ﬁlled with MIN tissue with no evidence of malignant tumors However, lactating glands revealed increased uptake compared to (Suppl. Figure S2b, c). Eight weeks post transplantation (w8), MIN background (Fig. 3a–c). During tumorigenesis, mean [ F]FDG and low amounts of invasive malignancy appeared in single uptake increased continuously (Fig. 3a, Table 1), while [ C]Chol lesions. However, by 11 weeks post transplantation (w11), tumors uptake rose initially from w3 to w7 and remained stable thereafter grew invasively through the border of the mammary fat pads (Fig. 3b, Table 1). [ Ga]RGD uptake surpassed background levels (Suppl. Figure S2 b, c). after only w10 (Fig. 3c, Table 1). None of the tracers reached Mean and maximal value analyses were performed based on signiﬁcant differences in mean uptake values compared to whole inguinal mammary fat pads. Both analyses revealed a trend lactating glands throughout the whole course of the study. towards higher [ F]FDG accumulation along the different steps of disease progression to IC and yielded high standard deviations Cluster analysis.[ F]FDG clustering was performed only in the (SDs) (Suppl. Figure S2d, e). Signiﬁcant differences appeared in ﬁrst study. The second study data served as a test data set, where maximal value analysis between the time points (5.8 ± 2.1%ID/cc thresholds deﬁning the cluster borders from the ﬁrst study were in w4; 9.6 ± 4.3%ID/cc in w8; 14.0 ± 5.8%ID/cc in w11, Suppl. Figure applied, and conﬁrmed previous clustering (Suppl. Figure S3). As a S2d, e). However, the PET images and ex vivo analysis revealed npj Breast Cancer (2022) 41 Published in partnership with the Breast Cancer Research Foundation 1234567890():,; J. Griessinger et al. Fig. 1 GMM analysis of [ F]FDG over time correlates with different tumor stages. a Summed histograms of all analyzed lesions of the single time points (w4: n = 16; w8: n = 15; w11: n = 13). Due to spillover effects from surrounding tissues (e.g., bladder), some lesions were excluded from the analysis. The three histograms were summed, and GMM analysis was performed on the summed data of all time points. b AIC and BIC for different numbers of Gaussian distributions, reaching its minimum at a sum of 5 Gaussian distributions. c Calculated thresholds from the 5 Gaussian mixture model were transferred to the summed histogram to separate the populations. d Applying the thresholds to the single time points demonstrates the appearance of the low uptake population in blue (< 6.7%ID/cc), the moderate uptake population in green (6.7–11.8%ID/cc), and the highest uptake population in red (> 11.8%ID/cc) for different time points. e Parametric maps of the lesions of two representative mice conﬁrm the appearance of the increased uptake populations (green and red) during tumorigenesis. f Autoradiography of different tumor slides and corresponding H&E veriﬁed the highest [ F]FDG uptake in solid tumor regions, whereas hyperplastic regions of the lesions showed the lowest uptake. The white scale bar indicates 5 mm. difference in both studies, tumor development was slower in the increased from w3 (3% of the total tumor volume) to w7 (16% second study. However, the cluster analysis reﬂected this fact of the total tumor volume) and remained relatively stable until accurately, showing a slower progression from low and moderate w13 (18% of total tumor volume) (Fig. 4d), correlating with the to high uptake populations (Suppl. Figure S3c–f). appearance of premalignant DCIS-like MIN regions (w7–w13) in 18 11 Similar to [ F]FDG clustering (Fig. 4a, b), cluster analysis of [ C] the histological description of the MIN-O model (Fig. 2a). In Chol also resulted in a sum of 4 Gaussian distributions with contrast, the population with the highest [ F]FDG uptake (> thresholds of 1.4%ID/cc, 2.4 %ID/cc, and 4.4%ID/cc (Fig. 4c, d). The 11.8%ID/cc) increased continuously from w3 (0% of total tumor population with the highest [ C]Chol uptake (>4.4%ID/cc) volume) until w13 (12% of total tumor volume) (Fig. 4b). Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 41 J. Griessinger et al. Fig. 2 Tracer accumulation during tumorigenesis of the MIN-O model compared to prelactating and lactating mammary glands. a H&E histology slides illustrating the development of the MIN-O lesions within the mammary glands compared to controls of a healthy and lactating gland. b Anatomical MRI and fused PET/MR images of a representative MIN-O mouse over time compared to pre- and lactating 18 11 68 glands for the investigated tracers [ F]FDG, [ C]Chol and [ Ga]RGD (white arrows indicate the region of the glands and MIN-O, respectively), and tracer uptake of all tracers increased during tumorigenesis over time. This increase in invasive tumor volume was also supported by Table 1. Results of mean value analysis. histology, which also identiﬁed further subtypes in w13 (Suppl. Figure S4) . 18 11 68 [ F]FDG [ C]Chol [ Ga]RGD 18 11 Both tracers, [ F]FDG and [ C]Chol, separated DCIS-like MIN and IC from the lactating glands using the corresponding Tissue %ID/cc [MBq] %ID/cc [MBq] %ID/cc [MBq] 18 11 population thresholds ([ F]FDG uptake > 6.7%ID/cc, [ C]Chol Muscle 1.0 ± 0.3 1.4 ± 0.4 0.5 ± 0.3 uptake > 4.4%ID/cc, Table 1). Pre-lactating glands 1.0 ± 0.1 1.3 ± 0.3 0.7 ± 0.2 Histological analysis supported these results and revealed Lactating glands 2.9 ± 0.2 1.7 ± 0.2 1.2 ± 0.7 changing histological patterns over time (Suppl. Figure S4): in Tumor (w3) 3.0 ± 0.9 2.1 ± 0.8 0.6 ± 0.2 w3, only the atypical hyperplastic type MIN tissue was observed. More pronounced higher-grade MIN appeared from w7 through Tumor (w7) 3.6 ± 0.9 2.9 ± 1.3 0.6 ± 0.2 w13. First signs of invasive growth were detectable in individual Tumor (w10) 4.4 ± 2.0 3.0 ± 1.0 0.7 ± 0.4 transplants as early as in w8 (study 1). In a second study, detection Tumor (w13) 3.7 ± 3.0 2.9 ± 1.0 1.4 ± 0.6 began in w10 (study 2), which involved the entire transplant by npj Breast Cancer (2022) 41 Published in partnership with the Breast Cancer Research Foundation J. Griessinger et al. Fig. 3 Mean value analysis of tracer uptake. Mean values (black bars) ± standard deviation, as well as single values, for every lesion (ﬁlled diamonds) of the investigated tracer were plotted over time and compared to prelactating (gray line) and lactating (black line) mammary 18 11 glands. a [ F]FDG uptake showed a steady increase in uptake from w3, starting at the level of the lactating glands to w13. b [ C]Chol showed a slight increase from w3 to w7 and a stable mean value from w7 until w13. c [ Ga]RGD uptake was observed in the range of the prelactating glands from w3 to w10 and increased from w10 to w13. A statistically signiﬁcant difference between MIN-O lesions and pre- or lactating glands was not observed for any of the tracers. w13. Within these large metaplastic tumors, cystic, eosinophilic, zones’ positive cells were also stronger than that in differentiation and glandular differentiation patterns appeared (Suppl. Figure S4). zones (Fig. 6d, e, j, k). 18 11 The speciﬁc uptake pattern of [ F]FDG and [ C]Chol within the The highest levels of GLUT1 signal were observed in same tumor differed (Fig. 4e). Autoradiography veriﬁed the adenocarcinomas (Fig. 6f, l) which also showed more positive highest [ F]FDG uptake in areas with invasive tumor growth cells than other zones on average (Fig. 6m, n; Mean ± SD = 91.0 ± within the tissue transplant sections, while the highest [ C]Chol 4.3% and 72.1 ± 6.2%, for 7-week and 10-week samples, respec- uptake appears correlated with premalignant MIN regions (Fig. 4f). tively). The same trend was also observed in 13-week MIN-O Clustering of [ Ga]RDG revealed a sum of only three uptake tissues (data not shown). populations with thresholds of 0.7%ID/cc and 1.6%ID/cc (Fig. 5a). This pattern of robust GLUT1 positivity, especially in adeno- Subtracting the background uptake in muscle and lactating carcinoma, correlates greatly with the results derived from [ F] glands (Table 1), only the highest uptake population remained. FDG uptake. Since the levels of GLUT1 are higher in adenocarci- This positive [ Ga]RGD population was only observed in large noma compared to the proliferation and differentiation zones, this tumors (Fig. 5a, b). difference can be correlated with tumor progression. In a similar fashion, Ki67 is also detected in a wide range of cells in 7-week and 10-week proliferation and differentiation zones (Fig. Immunohistochemical veriﬁcation 7m, n; Mean ± SD = 23.8 ± 9.5% and 64.6 ± 6.1%, for 7-week and To correlate the in vivo imaging data to the biological 10-week samples, respectively); whereas most cells in differentia- characteristics of the tissues, immunohistochemical analyses of tion zones were negative for this biomarker (Fig. 7m, n; Mean ± glucose transporter 1 (GLUT1) and proliferation (Ki67) were SD = 2.4 ± 1.4% and 26.0 ± 5.4%, for 7-week and 10-week samples, performed on 7-week and 10-week MIN-O tissues, which were respectively), constituting a signiﬁcant difference between the further classiﬁed into proliferation zones, differentiation zones, two zones at both 7 weeks (p = 0.0001, n = 6) and 10 weeks (p < and adenocarcinoma (Figs. 6, 7). Additional IHC included CD31- 0.0001, n = 6). and β -Integrin-staining for vessel density and neoangiogenesis, Notably, Ki67 expression remained consistently high in adeno- respectively (Suppl. Figure S5). carcinoma across these timepoints (Fig. 7m, n; Mean ± SD = Regarding GLUT1 expression, more cells in the proliferation 53.6 ± 5.7% and 53.7 ± 8.8%, for 7-week and 10-week samples, zones were positive than cells in differentiation zones at both respectively). 7 weeks (Fig. 6m; proliferation zones: Mean ± SD = 88.0 ± 3.8%; Interestingly enough, a distinctive pattern was observed in β3- differentiation zones: Mean ± SD = 54.2 ± 18.2%, p = 0.0002, n = integrin expression where the positive cells’ signals appeared 6) and 10 weeks (Fig. 6n; proliferation zones: Mean ± SD = 66.1 ± higher in adenocarcinoma while maintaining at relatively lower 8.0%; differentiation zones: Mean ± SD = 28.4 ± 8.0%, p < 0.0001, levels across all other zones (Suppl. Figure S5). In these n = 6). In addition, the levels of GLUT1 signal in proliferation adenocarcinoma areas, however, positive β3-integrin staining Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 41 J. Griessinger et al. was not only limited to capillary veins but also greatly detected in distribution and dense blood vessel system for MIN regions and tumor cells. Therefore, [ Ga]RGD uptake appeared to be mostly adenocarcinoma (Suppl. Figure S6). Our prior studies demon- mediated by the tumor cells themselves. strated the high level of vascularity in these lesions and the highly Addressing blood vessel density and formation within the disorganized pattern . A quantitative image analysis in MIN-O lesions, CD31 staining showed a dense but heterogenous tissues was further impaired by potential vasculogenic mimicry . npj Breast Cancer (2022) 41 Published in partnership with the Breast Cancer Research Foundation J. Griessinger et al. 18 11 18 Fig. 4 [ F]FDG and [ C]Chol clustering mismatch identiﬁes tumor stages. a The summed and ﬁtted histogram of [ F]FDG, as well as b the volumetric percentage of the uptake populations within the lesions for every time point, demonstrated a high amount of the low [ F]FDG uptake populations within the lesions in this study, including at later time points. The high uptake population (> 11.8%ID/cc) showed a slight increase over time up to 12% of the late-stage lesions in w13. c The summed histogram of all time points of [ C]Chol was analyzed as described for [ F]FDG, and the sum of four Gaussian distributions were identiﬁed as the best model. d The volumetric percentage of the uptake populations within the lesions for every time point demonstrated a lower amount of low [ C]Chol uptake populations compared to [ F]FDG. After a slight increase of the higher uptake populations (green, 2.4–4.4%ID/cc and red, > 4.4%ID/cc) from w3 to w7 and w10, both remained relatively stable until w13. e The parametric maps showed the appearance of the highest uptake populations at w7 in single tumors 11 18 for both tracers, staying stable over time for [ C]Chol and increasing for [ F]FDG. f The autoradiography results of representative tumor slides 18 11 demonstrated the highest [ F]FDG uptake in IC regions (red arrows), whereas the highest [ C]Chol uptake was observed in proliferating regions including DCIS-like high-grade MIN (white arrows) rather than invasive tumor regions. 68 68 Fig. 5 GMM analysis of [ Ga]RGD characterized IC regions. a The summed histogram of [ Ga]RGD uptake of all tumors and all time points was ﬁtted and analyzed as described above for [ F]FDG, resulting in a sum of 3 Gaussian distributions as the best model. The thresholds 0.7 and 1.6%ID/cc were applied to the histograms of the single time points. b Representative PET/MR images revealed the highest uptake population (red, > 1.6%ID/cc) in only large tumors. Patient data a medium time point (e.g., 7w and 10w), each transplant has morphologically identiﬁable components of both MIN and Histologically validated invasive tumor lesions of patients pre- invasive carcinoma. Thus, the model provides an experimentally sented high [ F]FDG uptake with an SUV mean > 7. The highest veriﬁed and reproducible basis for the study of comparable uptake in DCIS regions (SUV max) was below 1.6, a threshold we 26 human precancer progression to invasive breast cancer. These deﬁned in our previous work for less aggressive regions (Suppl. experiments represent biological proof of principle that can then Table 1, Fig. 8). be applied to the human disease. Damonte et al. demonstrated that invasive tumor components develop from the “differentiation zone” comprised of premalig- DISCUSSION nant low-grade more well-differentiated MIN . Detection of this This study was designed to demonstrate the potential of in vivo transition early in tumorigenesis could enable early preventive molecular imaging for determining the status of mammary interventions. The in vivo detection and differentiation of these neoplasias. Differentiation of pure NST, NST/DCIS, and DCIS regions within early MIN tissues was limited by the spatial without invasive carcinoma is essential for subsequent therapeutic resolution of the PET scanner (1.6 mm maximal achievable decisions . In the era of precision medicine and personalized resolution ) and, in this context, the partial volume effect. With treatment, this noninvasive approach should be applicable in the advent of next generation high-resolution, high-sensitivity PET human breast imaging and may provide an important new 31 scanners , this restriction of PET imaging could be overcome in diagnostic approach especially in the setting of active surveillance patients. In our study, the autoradiography (0.05 mm pixel length) of presumed pure DCIS . proved increased [ F]FDG accumulation in the MIN-derived We demonstrate here, using a mouse model of DCIS with tumors compared to premalignant MIN. While not the only consistent progression to invasive carcinoma, the MIN-O trans- relevant glucose transporter, GLUT1 IHC further corroborates plant model , that the precancer or MIN can be distinguished these ﬁndings. Furthermore, despite high levels of Ki67-expressing from invasive carcinoma using molecular PET/MR imaging and cells, invasive carcinoma appeared photopenic compared to MIN adequate analysis. Tumors in the MIN-O mouse model developed regions in [ C]Chol autoradiography. Thus, uptake mechanisms from experimentally deﬁned precancerous cells within the between MIN and IC might differ. transplanted MIN. The model has experimentally reproducible Due to the GMM-based voxelwise analysis, the PET imaging 18 11 68 and morphologically distinct stages DCIS-like MIN and IC. Taken at tracers [ F]FDG, [ C]Chol, and [ Ga]RGD demonstrated the Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 41 J. Griessinger et al. Fig. 6 Quantitative GLUT1 expression. The higher levels of GLUT1 positive cells are observed at proliferation zones and tumor areas. GLUT1 stained MIN-O tissues at 7-week a–f and 10-week g–l were analyzed for positive cell density m, n and the levels of GLUT1 d–f, j–l. Images were analyzed at proliferation zones a, d, g, j, differentiation zones b, e, h, k and tumor areas (indicated as AdenoCa; c, f, i, l. The color indicates the levels of GLUT1 d–f, j–l as brighter color as higher GLUT1 levels and dark color as lower GLUT1 levels. m, n Graphs show the measurement of GLUT1 positive cell densities (positive percentage) at proliferation zones, differentiation zones and tumor areas in m 7-week and n10-week MIN-O tissues. Bars represent mean value ± standard deviation. Scale bar indicates 100 µm. All images were adjusted to same magniﬁcation. 11,33 neoplastic variations within each whole heterogeneous transplant target . In the MIN-O mouse model, in vivo imaging revealed lesion at single time points. Temporal changes were documented increased [ Ga]RGD uptake only in late-stage invasively growing by repeated studies in the same animal providing a chronological tumors, suggesting a late angiogenic switch during the develop- map of tumor development. Each lesion could be subsequently ment of invasive carcinoma. However β -integrin IHC primarily co-registered and veriﬁed using autoradiography and histology. showed overexpression on tumor cells of invasively growing This noninvasive imaging approach decoded the intra-lesional tumors rather than on blood vessels. Therefore, this angiogenic pattern of the mammary neoplastic stages. In particular, [ F]FDG switch could not be veriﬁed. In contrast, and supported by β - and [ C]Chol distinguished between premalignant MIN and IC, integrin staining, accumulation of [ Ga]RGD was only observed in which was directly supported by autoradiography. These lesions 68 invasive tumors, making [ Ga]RGD a potentially important marker also proved to differ from the physiologically increased metabo- for IC-development in this mouse model. lism and proliferation of prelactating and lactating mammary The presence of activated α β -integrin on the tumor cells V 3 gland controls. themselves, especially in breast cancer, is often associated with a [ Ga]RGD-PET has been discussed as a promising imaging 34–36 metastatic phenotype . This speciﬁc MIN-O line has a high modality for the investigation of breast cancer, because the target incidence of lung metastasis at later time-points; α β -integrin V 3 of the RGD peptide, the activated α β -integrin, is overexpressed V 3 18 expression could be an early step in metastasis . However, on newly generated blood vessels during neoangiogenesis . examining the lungs of our mice at w13 did not reveal metastases. α β -integrin has also been discussed as a potential therapeutic V 3 npj Breast Cancer (2022) 41 Published in partnership with the Breast Cancer Research Foundation J. Griessinger et al. Fig. 7 Quantitative Ki67 expression. Proliferation zones and tumor areas are highly proliferative compared with differentiation zones. Ki67 stained MIN-O tissues at 7-week a–f and 10-week g–l were analyzed for cell proliferation m, n. Images were analyzed at proliferation zone a, d, g, j, differentiation zones b, e, h, k and tumor areas (indicated as AdenoCa; c, f, i, l. The analysis for Ki67 positive cells were performed on QuPath by identifying nuclear Ki67 signals. Images shown as “PosNeg” indicate red cells as Ki67 positive and blue cells as Ki67 negative. The densities of Ki67 positive cells (positive percentage) were analyzed at proliferation zones, differentiation zones and tumor areas in m 7-week and n 10-week MIN-O tissues. Bars represent mean value ± standard deviation. m Scale bar indicates 100 µm. All images were adjusted to same magniﬁcation. Nevertheless, the time course of α β -integrin expression on This agreed with both our clinical results and the literature . The V 3 tumor cells and metastasis formation merits further study. additional information of higher [ C]Chol uptake at the prolifera- The timeline of occurrence of invasive carcinoma and asso- tion zone and DCIS-like high-grade MIN could translate into ciated DCIS is controversial. Mathematical modeling of DCIS and veriﬁcation of DCIS in the background of persistent breast tissue. associated invasive carcinoma origins and development based on This however could not be veriﬁed because the clinical data were empirical marker analyses suggest that they arise simultaneously not available for our patients. Further, Contractor et al. reported 11 41 and grow in parallel, as opposed to the prior assumption that DCIS [ C]Chol uptake in ER-positive breast cancer . Clearly, differences expands ﬁrst, and a subsequent focus of invasion evolves as a in tracer uptake and the underlying physiological processes in 37,38 subclone of the DCIS . Intriguingly this is supported by single- choline and glucose metabolism need to be further investigated. cell sequence analysis of DCIS and adjacent invasive cancers Characterization of the tumorigenesis of mammary cancer which display identical subclones in both compartments . In all originating from transplanted atypical hyperplastic outgrowths 11 18 its stages, the MIN-O model supports the parallel pathway theory was proven to be feasible using the radiotracers [ C]Chol, [ F] FDG, and [ Ga]RGD. Advanced imaging analysis distinguished for this model. That is, all information for the development of the 18–21,38,40 malignant disease is already encoded in the MIN stages . intra-transplant regions of premalignant DCIS-like high-grade MIN 11 18 A direct application of the MIN-O model to clinical studies from IC by [ C]Chol and [ F]FDG. Supported by β - integrin IHC, 18 68 remains debatable. [ F]FDG detected the more aggressive IC [ Ga]RGD-PET imaging indicated the progression from MIN to regions as compared to regions of MIN and normal tissue growth. adenocarcinoma, suggesting further correlations with a metastatic Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 41 J. Griessinger et al. Fig. 8 Representative images from patients. (Patient 1) A 48-year-old with biopsy-proven invasive ductal carcinoma (asterisks) in the right breast with extensive DCIS (arrows), which was proven in ﬁnal histopathology. Axial subtraction image of contrast-enhanced dynamic breast MRI in the upper half shows the intensively enhancing lesion (asterisk) with a linear non-mass-enhancement (arrow) towards the nipple. In the lower half, axial PET shows the high SUV of the invasive tumor (asterisk; SUV mean 7.39) and low SUV in the localization of DCIS (arrow; SUV max 1.27). (Patient 2) A 46-year-old with biopsy-proven invasive ductal carcinoma (asterisks) in the left breast with associated DCIS (arrows), which was proven in ﬁnal histopathology. Axial subtraction image of contrast-enhanced dynamic breast MRI in the upper half shows the intensively enhancing lesion (asterisk) with a linear non-mass-enhancement (arrow) towards the nipple. In the lower half, axial PET shows the high SUV of the invasive tumor (asterisk; SUV mean 7.39) and the low SUV in the localization of the DCIS (arrow; SUV max 1.27). breast cancer phenotype. Our results also support the hypothesis Table 2. Tracer uptake times and injected doses. of parallel development of IC and DCIS in this tumor model. Most importantly, our study shows that molecular imaging enables a Tracer IA Uptake time TP PET EM PET TX MRI localized differentiation between premalignant disease and invasive carcinoma. Further application of preclinical therapeutic [MBq] [min] [weeks] [min] Anatomy studies and translation of the imaging protocols to clinical S1: [ F] 13 ± 2 66 ± 3 4,8,11 10 Yes Yes investigations can provide a better understanding of DCIS FDG development and provide an approach that reduces overtreat- S2. [ F] 13 ± 1 60 3,7,10,13 10 Yes Yes ment. We hope that the lessons learned from our in vivo FDG experiments with a model system of cancer progression will guide S2: [ C] 13 ± 1 40 3,7,10,13 10 Yes Yes and stimulate similar studies of human breast cancer. Chol S2: [ Ga] 13 ± 1 70 3,7,10,13 10 Yes Yes RGD METHODS Mice S1 Metabolic alterations during tumor development, S2 Multiparametric All experiments were performed on female FVB/N mice (Charles River characterization of lesion stages compared to the lactating gland, IA Laboratories, Sulzfeld, Germany). The MIN-O line was generated by injected activity, TP time points, EM emission, TX transmission. Data are transplantation of MIN lesions to create transplantable outgrowth lines displayed as mean ± standard deviation, where applicable. (MIN-O) in the mammary glands of FVB/N-Tg(MMTV-PyVT)634Mul/J mice at the University of California, Davis, USA. Frozen MIN-O (line D) tissue was transferred to our lab for implantation into gland-cleared mammary fat attenuation correction was performed. Subsequently, the animals were pads of FVB/N mice and subsequent serial transplantation. Both inguinal transferred on the same bed to the MRI scanner. Three-dimensional T2- mammary fat pads of the animals were cleared from the developing gland weighted turbo spin-echo MRI (TSE; TR 2500 ms, TE 202 ms, voxel size structure, and tissue was transplanted on both sides when mice were 0.27 × 0.27 × 0.27 mm ) provided anatomical references. Table 2 sum- 18,42 3 weeks old, as previously described . All animals were housed under marizes relevant study details. standardized environmental conditions (22 ± 2 °C room temperature, 55 ± All experiments were performed according to animal use and care 10% relative humidity, and 12 h light-dark phases) with free access to food protocols approved by local authorities (regional council Tübingen). and water. Mice developed palpable tumors at 13–16 weeks of age (10–13 weeks post transplantation) in the transplanted mammary glands. Metabolic alterations during tumor development. Tissue metabolism Tumor development was monitored from week 3 through week 13 post during tumor development (n = 10 mice, bearing 20 transplanted lesions) transplantation. was investigated using the combined PET/MRI protocol 4, 8, and 11 weeks after transplantation (w4, w8, and w11). As one mouse was sacriﬁced in w4 and 2 mice in w8 for ex vivo analyses, 9 and 7 mice were measured in w8 Preclinical in vivo imaging and w11, respectively. The remaining 7 mice were sacriﬁced after the last In vivo studies were performed in a sequential PET/MRI setup using a measurement in w11, and lesions were excised for ex vivo analyses. dedicated 7 T small-animal MR scanner (ClinScan, Bruker BioSpin GmbH, Ettlingen, Germany) and a small-animal PET scanner (Inveon dedicated 27,43 18 Multiparametric characterization of lesion stages compared to the lactating PET, Siemens Healthcare, Knoxville, TN, USA) .[ F]FDG was synthesized gland. To account for an additional measurement time point and tumor 11 68 according to our marketing license, [ C]Chol and [ Ga]RGD followed our growth, the imaging time points were adjusted. Five animals in weeks 3, 7, 25,44 published procedures . Animals were anesthetized with 1–2% isoﬂur- and 10 post transplantation (w3, w7, and w10) and 4 animals in week 13 ane evaporated in breathing air. PET tracers were intravenously injected. post transplantation (w13) were measured on temperature-controlled After 10 min PET emission scan, a 13 min PET transmission scan for npj Breast Cancer (2022) 41 Published in partnership with the Breast Cancer Research Foundation J. Griessinger et al. animal beds. All measurements were performed in a sequential PET/MRI Beyond this analysis, data were analyzed using a voxel-based analysis 18 11 26 setup after intravenous injection of 13 ± 1 MBq of [ F]FDG, [ C]Chol, or approach as previously described . Brieﬂy, histograms of the voxel values [ Ga]RGD (Radiopharmacy, University Hospital Tuebingen, Tuebingen, of PET data were calculated for the sum of all lesions within each time Germany). All tracers were measured in w3, w7, w10, and w13 within one point and for the sum of all time points within the study. A Gaussian 11 18 week. Of note, [ C]Methionine and [ F]FMISO have also been measured mixture model (GMM) was applied to cluster multiple uptake populations but did not provide further information. of the tracers over the course of the study during tumor development An additional ﬁve mice were measured only with [ F]FDG-PET/MRI. Two using Akaike information criterion (AIC) and Bayesian information criterion of these ﬁve mice were sacriﬁced in w3 and w7 and one in w10, and (BIC) to select the number of clusters (MATLAB, MathWorks, Natick, USA). lesions were excised for ex vivo analysis. One additional mouse from the The intersection between adjacent clusters deﬁned the respective cluster multi-tracer measurements was sacriﬁced in week 10 for ex vivo analysis, boundaries. resulting in 2 mice per time point for ex vivo analyses. One mouse died 68 Quantitative image analysis was performed digitized slides using the during w13 before [ Ga]RGD examination; therefore, an additional mouse 68 QuPath software (Version 0.2.3 ). The analysis for GLUT1 positive cells was from the same transplantation was measured only with [ Ga]RGD during performed by identifying its signals in cytoplasmic/membrane. Images are this week. All mice were sacriﬁced in w13, and lesions were excised for shown as “GLUT1 levels” indicate the levels of GLUT1 as a heatmap with jet ex vivo analyses. color scale. The GLUT1 positive cells (positive percentage) were analyzed at Furthermore, an additional six non-tumor-bearing mice (two for each tracer) were measured on day 16 ± 1 of their pregnancy and day 5 ± 1 of growth zones, differentiation zones and tumor areas in (m)7-week and (n) lactation to assess pre- and lactating mammary glands. Following the last 10-week MIN-O tissues. The analysis for Ki67 positive cells were performed measurement, one mouse per tracer was sacriﬁced, and the lactating by identifying nuclear Ki67 signals. Images shown as “PosNeg” indicate red mammary gland was excised for ex vivo analyses. The remaining mouse cells as Ki67 positive and blue cells as Ki67 negative. The densities of Ki67 nursed the offspring of both mice. positive cells (positive percentage) were analyzed at growth zones, differentiation zones and tumor areas in (m)7-week and (n)10-week MIN- O tissues. Ex vivo methods β -Integrin staining was visualized similar to GLUT1. A quantitative Following the last in vivo measurement, mice were sacriﬁced, and lesions evaluation of the different zones was omitted due to the positive staining were excised and prepared for whole-mount staining, histology, and or of the adenocarcinoma cells. autoradiography. For data correlation with [ F]FDG autoradiography, performed during the study time points, an additional two mice were 11 Patient data injected with 13 ± 1 MBq [ C]Chol in w8 to perform an autoradiography Patient data were obtained as part of an IIT trial (German Clinical Trials experiment after 60 min of uptake. Register, DRKS00013891) after approval by the responsible Ethics Committee of the Medical Faculty of the Eberhard-Karls-University and Autoradiography. For autoradiography, tumors were embedded in Tissue- the University Hospital Tübingen and after patients provided written Tek O.C.T. compound (Sakura Finetek, Torrance, CA, USA) and frozen at informed consent. Data were retrospectively analyzed to show two −20 °C. Every 200 μm, a 20 μm section was cut with a cryostat (Leica examples of imaging characteristics in tumors with extensive intraductal Microsystems, Wetzlar, Germany) at −19 °C. A storage phosphor screen was placed on the slices and read out after an exposure time of 10 half- components of an invasive tumor. The patients presented with grade 3 lives of the respective tracer with a pixel size of 50 μm using a STORM unifocal and multifocal invasive breast cancer of no special type with Phosphor-Imager (Molecular Dynamics, Sunnyvale, CA, USA). Tissue slices associated high-grade DCIS, ER/PR neg, both IRS 0%, and lymph-node were then stained with hematoxylin and eosin (H&E), and whole-slide involvement. The two breast cancer patients received multimodal PET/MR, images were digitized using a digital slide scanner (NanoZoomer-XR 18 including 244 and 247 MBq of [ F]FDG and 0.1 mmol/kg Gd-based C12000, Hamamatsu Photonics K.K., Hamamatsu-City, Japan). For normal- contrast agent (6 and 8 ml gadubutrol, respectively). To determine tracer ization, autoradiography was analyzed as tumor-to-muscle-ratios (TMR), uptake, ROIs were drawn on manually identiﬁed lesions. After surgical dividing the whole autoradiography plate of each mouse by the mean resection of the lesions, the tissue was histologically characterized. value of the muscle samples on the plate (ImageJ; National Institute of Health, Bethesda, USA ). Statistical analysis Histology. For histology, tumors were ﬁxed in 4.5% formalin (SAV LP Mean and maximal value analysis was performed based on the uptake in % GmbH, Flintsbach am Inn, Germany) or zinc (IHC Zinc-Fixative, BD ID/cc (mean/max value ± standard deviation (SD)). For statistical analysis of Biosciences, Franklin Lakes, USA) and embedded in parafﬁn (Paraplast® PET data to compare the tracer uptake of different time points, Tukey- Embedding Media, McCormick Scientiﬁc, Leica Microsystems, Wetzlar, Kramer tests were performed (JMP, SAS Institute, Cary, USA). To compare Germany). Formalin-ﬁxed sections were cut at 4 μm, deparafﬁnized and tracer uptake of the time points in tumor development with the control stained with either H&E or with the respective antibodies for immunohis- tissue of pre- and lactating mammary glands, Dunnett’s tests, including tochemistry. Slides were stained using an automatic immunostainer Bonferroni correction, were performed (JMP). (Discovery XT, Ventana Medical Systems, Inc., Tucson, USA) according to Statistical comparison of quantitative image analysis was performed the manufacturer’s standard protocols with antibodies against CD31 using a one-way ANOVA with Tukey-correction for multiple comparisons (Abcam Inc., Cambridge, USA), β -Integrin (Abcam Inc.), glucose transporter (GraphPad Prism V6). 1 (GLUT1, Abcam Inc.) and Ki67 (ThermoFisher Scientiﬁc, Waltham, USA). Images were acquired under a microscope (Axio Imager A1, Carl Zeiss AG, Oberkochen, Germany) using a coupled digital camera (ProgRes® C10plus, Reporting summary JenOptik, Jena, Germany) and the software ImageAccess (Version 6, Imagic Further information on research design is available in the Nature Research Bildverarbeitungs AG, Glattbrugg, Switzerland) or extracted from digitized Reporting Summary linked to this article. whole slide imaging (Nanozoomer, Hamamatsu) using the corresponding software. DATA AVAILABILITY Data analysis All relevant data are presented in the manuscript and supplemental material. For subsequent data re-use a Data Usage and Access Committee (DUAC) will provide Preclinical PET data were reconstructed using the ordered-subsets access to research-relevant data (anonymized) for research purposes after submitting expectation maximization algorithm with 3D post reconstruction a reasonable data usage request to the corresponding author. (OSEM3D) (Inveon Acquisition Workplace, Siemens Healthcare). For in vivo data analysis, Inveon Research Workplace (IRW, Siemens Healthcare) was used. Corresponding PET and MR images were fused, volumes of CODE AVAILABILITY interest (VOIs) covering the entire mammary fat pads were drawn based on the anatomical MRI data and the mean and maximal uptake values (%ID/ Matlab scripts are available after submitting a reasonable code usage request to the cc) of the detected lesions were calculated. corresponding author. Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 41 J. Griessinger et al. Received: 18 January 2021; Accepted: 1 February 2022; 27. Kemp, B. J. et al. NEMA NU 2-2007 performance measurements of the Siemens Inveon preclinical small animal PET system. Phys. Med. Biol. 54, 2359 (2009). 28. Lim, D. et al. Angiogenesis and vasculogenic mimicry as therapeutic targets in ovarian cancer. BMB Rep. 53, 291–298 (2020). 29. Esserman, L. J. et al. The evolution of our understanding of the biology of cancer is the key to avoiding overdiagnosis and overtreatment. Cancer Epidemiol. Bio- REFERENCES markers Prev. 2020. 1. Esserman, L. J. et al. Use of molecular tools to identify patients with indolent 30. Kanbayashi, C. et al. The international collaboration of active surveillance trials for breast cancers with ultralow risk over 2 decades. JAMA Oncol. 3, 1503–1510 low-risk. DCIS (LORIS, LORD, COMET, LORETTA) 37, TPS603 (2019). (2017). 31. Badawi, R. D. et al. First human imaging studies with the EXPLORER total-body 2. Swanton, C. Intratumor heterogeneity: evolution through space and time. Cancer PET scanner. J. Nucl. Med. 60, 299–303 (2019). Res. 72, 4875–4882 (2012). 32. Haubner, R. et al. Glycosylated RGD-containing peptides: tracer for tumor tar- 3. Kreienberg, R. et al. Interdisciplinary GoR level III guidelines for the diagnosis, geting and angiogenesis imaging with improved biokinetics. J. Nucl. Med. 42, therapy and follow-up care of breast cancer: short version–AWMF Registry No.: 326–336 (2001). 032-045OL AWMF-Register-Nummer: 032-045OL–Kurzversion 3.0, Juli 2012. 33. Yoon, H.-J. et al. Correlation of breast cancer subtypes, based on estrogen Geburtshilfe und Frauenheilkd. 73, 556 (2013). receptor, progesterone receptor, and HER2, with functional imaging parameters 4. Ignatiadis, M. & Sotiriou, C. Understanding the molecular basis of histologic from 68Ga-RGD PET/CT and 18F-FDG PET/CT. Eur. J. Nucl. Med. Mol. Imaging 41, grade. Pathobiology 75, 104–111 (2008). 1534–1543 (2014). 5. Heusch, P. et al. Thoracic staging in lung cancer: prospective comparison of 18F- 34. Felding-Habermann, B. et al. Integrin activation controls metastasis in human FDG PET/MR imaging and 18F-FDG PET/CT. J. Nucl. Med. 55, 373–378 (2014). breast cancer. Proc. Natl Acad. Sci. USA 98, 1853–1858 (2001). 6. Afshar-Oromieh, A. et al. Comparison of PET imaging with a 68Ga-labelled PSMA 35. Teti, A., Migliaccio, S. & Baron, R. The role of the alphaVbeta3 integrin in the ligand and 18F-choline-based PET/CT for the diagnosis of recurrent prostate development of osteolytic bone metastases: a pharmacological target for alter- cancer. Eur. J. Nucl. Med. Mol. Imaging 41,11–20 (2014). native therapy? Calcif. Tissue Int. 71, 293–299 (2002). 7. Collins, C. D. PET in lymphoma. Cancer Imaging 6, S63 (2006). 36. Rolli, M. et al. Activated integrin αvβ3 cooperates with metalloproteinase MMP-9 8. Jansson, T. et al. Positron emission tomography studies in patients with locally in regulating migration of metastatic breast cancer cells. Proc. Natl Acad. Sci. USA advanced and/or metastatic breast cancer: a method for early therapy evalua- 100, 9482–9487 (2003). tion? J. Clin. Oncol. 13, 1470–1477 (1995). 37. Sontag, L. & Axelrod, D. E. Evaluation of pathways for progression of hetero- 9. Avril, N. et al. Glucose metabolism of breast cancer assessed by 18F-FDG PET: geneous breast tumors. J. Theor. Biol. 232, 179–189 (2005). histologic and immunohistochemical tissue analysis. J. Nucl. Med. 42,9–16 (2001). 38. Kuerer, H. M. et al. Ductal carcinoma in situ: state of the science and roadmap to 10. Buck, A. K., Schirrmeister, H., Mattfeldt, T. & Reske, S. N. Biological characterisation advance the ﬁeld. J. Clin. Oncol. 27, 279–288 (2009). of breast cancer by means of PET. Eur. J. Nucl. Med. Mol. Imaging 31, S80–S87 39. Casasent, A. K., Edgerton, M. & Navin, N. E. Genome evolution in ductal carcinoma (2004). in situ: invasion of the clones. J. Pathol. 241, 208–218 (2017). 11. Beer, A. J. et al. Patterns of αvβ3 expression in primary and metastatic human 40. Cardiff, R. D. & Borowsky, A. D. Precancer: sequentially acquired or pre- breast cancer as shown by 18F-Galacto-RGD PET. J. Nucl. Med. 49, 255–259 (2008). determined? Toxicol. Pathol. 38, 171–179 (2010). 12. Cheng, J. et al. 18F-ﬂuoromisonidazole PET/CT: a potential tool for predicting 41. Contractor, K. B. et al. Biological basis of [(1)(1)C]choline-positron emission primary endocrine therapy resistance in breast cancer. J. Nucl. Med. 54, 333–340 tomography in patients with breast cancer: comparison with [(1)(8)F]ﬂuor- (2013). othymidine positron emission tomography. Nucl. Med. Commun. 32, 997–1004 13. Henry, K. E., Ulaner, G. A. & Lewis, J. S. Human epidermal growth factor receptor (2011). 2-targeted PET/single-photon emission computed tomography imaging of breast 42. Young, L. J. The cleared mammary fat pad and the transplantation of mammary cancer: noninvasive measurement of a biomarker integral to tumor treatment gland morphological structures and cells, in Methods in mammary gland biology and prognosis. PET Clin. 12, 269–288 (2017). and breast cancer research. 2000, Springer. p. 67–74. 14. Smyczek-Gargya, B. et al. PET with [18F]ﬂuorothymidine for imaging of primary 43. Mannheim, J. G. et al. Quantiﬁcation accuracy and partial volume effect in breast cancer: a pilot study. Eur. J. Nucl. Med. Mol. Imaging 31, 720–724 (2004). dependence of the attenuation correction of a state-of-the-art small animal PET 15. Wong, H. et al. Presence of an in situ component is associated with reduced scanner. Phys. Med. Biol. 57, 3981–3993 (2012). biological aggressiveness of size-matched invasive breast cancer. Br. J. Cancer 44. Reischl, G. et al. Highly efﬁcient automated synthesis of [(11)C]choline for multi 102, 1391–1396 (2010). dose utilization. Appl. Radiat. Isot. 60, 835–838 (2004). 16. Dieterich, M. et al. Accompanying DCIS in breast cancer patients with invasive 45. Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of ductal carcinoma is predictive of improved local recurrence-free survival. Breast image analysis. Nat. Methods 9, 671–675 (2012). 23, 346–351 (2014). 46. Bankhead, P. et al. QuPath: Open source software for digital pathology image 17. Kole, A. J. et al. Overall survival is improved when DCIS accompanies invasive analysis. Sci. Rep. 7, 16878 (2017). breast cancer. Sci. Rep. 9, 9934 (2019). 18. Maglione, J. E. et al. Polyomavirus middle T–induced mammary intraepithelial neoplasia outgrowths: single origin, divergent evolution, and multiple outcomes. ACKNOWLEDGEMENTS Mol. Cancer Therapeutics 3, 941 (2004). We would like to thank Maren Harant, Ramona Stumm, Funda Cay, Natalie Mucha, 19. Damonte, P. et al. Mammary carcinoma behavior is programmed in the precancer Daniel Bukala, Neil Hubbard, and Sandro Aidone for their excellent technical stem cell. Breast Cancer Res. 10, R50 (2008). assistance. Further, parts of this work were included in the doctoral thesis of Jennifer 20. Namba, R. et al. Heterogeneity of mammary lesions represent molecular differ- Griessinger, nee Schmitz. Parts of this work were funded by fortüne grant 2223-0-0, ences. BMC Cancer 6, 275 (2006). the Swiss Werner Siemens Foundation, the Deutsche Forschungsgemeinschaft (DFG, 21. Namba, R. et al. Molecular characterization of the transition to malignancy in a genetically engineered mouse-based model of ductal carcinoma in situ. Mol. German Research Foundation) under Germany’s Excellence Strategy—EXC 2180— Cancer Res. 2, 453–463 (2004). 390900677 and Institutional Strategy of the University of Tübingen (Deutsche 22. Abbey, C. K. et al. Preclinical imaging of mammary intraepithelial neoplasia with Forschungsgemeinschaft, ZUK 63). positron emission tomography. J. Mammary Gland Biol. Neoplasia 11, 137–149 (2006). 23. Abbey, C. K. et al. In vivo positron-emission tomography imaging of progression AUTHOR CONTRIBUTIONS and transformation in a mouse model of mammary neoplasia. Proc. Natl Acad. Sci. J.G., Q.C., B.J.P., R.D.C., A.M.S. planned and designed the experiment. J.G., A.K., Q.C., USA 101, 11438–11443 (2004). U.K. performed the experiments and processed the samples, J.G., J.S., Q.C., A.N.D., 24. Namba, R. et al. Rapamycin inhibits growth of premalignant and malignant H.M., L.Q.M., U.K., O.H.A., A.D.B., R.D.C., B.J.P., A.M.S. analyzed and interpreted the data, mammary lesions in a mouse model of ductal carcinoma in situ. Clin. Cancer Res. J.C., G.B., G.R. generated and controlled the radioactive tracers, H.P. collected and 12, 2613–2621 (2006). analyzed the clinical data. J.G., H.P., R.D.C., A.M.S. wrote the manuscript; all authors 25. Knetsch, P. A. et al. [68Ga]NODAGA-RGD for imaging αvβ3 integrin expression. revised the manuscript and approved the submitted version. Eur. J. Nucl. Med. Mol. Imaging 38, 1303–1312 (2011). 26. Schmitz, J. et al. Decoding intratumoral heterogeneity of breast cancer by mul- tiparametric in vivo imaging: a translational study. Cancer Res. 76, 5512–5522 FUNDING (2016). Open Access funding enabled and organized by Projekt DEAL. npj Breast Cancer (2022) 41 Published in partnership with the Breast Cancer Research Foundation J. Griessinger et al. COMPETING INTERESTS Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, The authors declare no competing interests. adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party ADDITIONAL INFORMATION material in this article are included in the article’s Creative Commons license, unless Supplementary information The online version contains supplementary material indicated otherwise in a credit line to the material. If material is not included in the available at https://doi.org/10.1038/s41523-022-00398-x. article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly Correspondence and requests for materials should be addressed to Andreas M. from the copyright holder. To view a copy of this license, visit http://creativecommons. Schmid. org/licenses/by/4.0/. Reprints and permission information is available at http://www.nature.com/ reprints © The Author(s) 2022 Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 41
npj Breast Cancer – Springer Journals
Published: Mar 24, 2022
Access the full text.
Sign up today, get DeepDyve free for 14 days.