Keywords Artificial intelligence - heterogeneity - molecular imaging - musculoskeletal - precision
medicine - radiomics
Introduction
The number of new cancer cases in 2011–2015 was 439.2/100,000 persons/year, with approximately
163.5 cancer-related deaths/100,000 persons/year.[1 ] Cancers arise from complex biochemical cellular processes secondary to alterations
in normal DNA, often resulting in uncontrolled rapid cellular proliferation. Tumor
biomarkers are essential in the diagnosis, risk-stratification, and treatment planning
of tumors. With the continual growing emphasis on genomics, proteomics, and radiomics,
as well as advances in molecular imaging, personalized precision medicine is becoming
a tangible reality. This manuscript aims to provide an overview of molecular imaging
for musculoskeletal (MSK) malignancies, highlighting the role it may play in the era
of precision medicine.
Tumor Heterogeneity, Genomic Biomarkers, and Molecular Imaging
Cancers consist of a heterogeneous collection of cell with various mutations, leading
to different biologic properties, including degrees of differentiation and growth
rate.[2 ],[3 ] This heterogeneity serves as a strong internal mechanism for tumor cells to escape
various oncologic treatments. Cancer cell heterogeneity can be categorized as intertumoral
and intratumoral. Intertumoral heterogeneity alludes to various biological properties
among different lesions of an identical malignancy. Intertumoral heterogeneity arises
from a combination of intrinsic and extrinsic mechanisms, including genetic and epigenetic
mutations and influences of the tumor microenvironment, causing varying biology of
the same tumor type between patients or even different lesions within the same patient.[4 ] Intratumoral heterogeneity refers to the microheterogeneity within a tumor, in part
secondary to imperfect rapid DNA replication in rapidly growing cancers. This leads
to a diverse population of cancer cell types within a single lesion, creating difficulties
in interpreting limited tissue sampling of a malignancy, such as a biopsy, and determining
appropriate therapeutic management.[5 ],[6 ]
Genetic mutations in tumors can consist of oncogenes (such as c-myc, fos, Ha-ras,
Ki-ras, sis, met, SAS MFH, and MDM2), tumor suppressor genes (such as p53, Rb, NF1,
and APC), and tumor-specific translocations (such as CHOP-FUS [TLS], EWS-FLI1, EWS-ATF1,
SYT-SSX, and PAX3-FKHR).[3 ],[7 ],[8 ] Traditional medical management of tumors typically involves obtaining a single sample
of a tumor and determining the appropriate therapeutic option from that encapsulating
diagnosis. Precision medicine aims to capture both the inter- and intratumoral heterogeneity
within a patient to create a personalized treatment plan. Molecular imaging noninvasively
images the complex biochemical and genetic processes of cancers. This imaging consists
of various physiologic imaging techniques targeting components such as peptides, antibodies,
proteins, affibodies, aptamers, and nanoparticles, predominantly in the field of nuclear
medicine, as well as analysis of quantitative data from cross-sectional imaging, such
as computed tomography (CT), magnetic resonance imaging (MRI), and ultrasound (US).
Utilizing various imaging techniques, molecular imaging provides a realistic method
to better quantify tumor heterogeneity throughout a patient.[9 ],[10 ],[11 ] Molecular imaging not only provides an insight into initial personalized cancer
treatment decisions, but also allows for continual monitoring during treatment. This
may lead to the detection of new cancer mutations during treatment, which could prompt
changes in therapy before other signs of tumor progression.[12 ],[13 ],[14 ],[15 ] With continuing improvements in molecular imaging techniques and devices, recognition
of new genetic and molecular targets, and new methods of analyzing and quantifying
data with artificial intelligence, there is an increasing role of molecular imaging
in the diagnosis and treatment of MSK malignancies [Figure 1 ].
Figure 1 From omics to molecular imaging and precision medicine
Physiologic Imaging
Bone scintigraphy
Nuclear medicine bone scintigraphy, most commonly with the use of99m Tc-methylene
diphosphonate (MDP), is a functional measurement of bone metabolism. It can play a
significant role in the evaluation of osseous metastases and cancer staging, and help
distinguish metabolically-inactive treated bone metastases from active disease. The
specificity, sensitivity, and accuracy for bone scintigraphy for the detection of
osseous metastases are 80.9%–96%, 67%–95.2%, and 60%–80.3%, respectively.[16 ] Bone scintigraphy can be performed with either a singlestatic phase to identify
regions of bone with high osteoblast activity, or as a dynamic threephase study, with
additional perfusion and blood pool phases to help distinguish inflammatory conditions
and changes in blood supply. With a high sensitivity, bone scans are useful in identifying
new metastatic lesions. However, the study is limited due to radiotracer uptake up
by a variety of other disease processes, including metabolic bone diseases, infections,
traumatic injury, and inflammatory conditions.[17 ],[18 ],[19 ],[20 ]
Single-photon emission tomography
Single-photon emission computed tomography (SPECT) scans are spatial three-dimensional
acquisitions of radionuclides. With multiplanar reconstruction, SPECT allows for better
contrast resolution and improvement lesion localization. In addition, SPECT can be
fused with CT to allow for concurrent anatomical and functional imaging, resulting
in improved specificity, sensitivity, and spatial resolution for MSK malignancies.[21 ],[22 ],[23 ] In particular, SPECT-CT has been shown to reduce equivocal interpretations compared
to SPECT or planar scintigraphy in MSK malignancies.[21 ],[24 ],[25 ],[26 ],[27 ],[28 ]
Positron emission tomography
The development and advances in positron emission tomography (PET) have revolutionized
functional imaging. With the use 18 F-fluorodeoxyglucose (18 F-FDG) to evaluate tumor metabolism, and various other radiopharmaceuticals targeting
specific molecular targets, PET has now plays a big role in the accurate staging and
monitoring of MSK malignancies, and can also serve as a predictor for treatment outcomes
[Figure 2 ], [Figure 3 ], [Figure 4 ].[29 ],[30 ],[31 ],[32 ]
Figure 2 A 44-year-old man with carcinoma of unknown primary. The bone 99m Tc-methylene diphosphonate scintigraphy demonstrated several skeletal lesions throughout
the body, 99m Tc-prostate-specific membrane antigen scintigraphy and 18 F-fluorodeoxyglucose positron emission tomography images showed avid lesions only
in the pelvis, and 99m Tc-octreotide scintigraphy demonstrated no activity, highlighting the intertumoral
heterogeneity
Figure 3 A 29-year-old man with poorly differentiated neuroendocrine tumor (Ki-67 = 28%).
99m Tc-octreotide scintigraphy and post-177 Lu-DOTATATE therapy images showed intense uptake within the skeletal lesions, predicting
a good response to 177 Lu-DOTATATE therapy in patients with somatostatin-expressing neuroendocrine tumors.
However, 18 F-fluorodeoxyglucose positron emission tomography-computed tomography images demonstrated
numerous 18 F-fluorodeoxyglucose-avid lesions throughout the skeleton and marrow, representing
a poor prognosis
Figure 4 An 8-year-old boy with Stage IV neuroblastoma. 18 F-fluorodeoxyglucose positron emission tomography-computed tomography images demonstrated
faint-18 F-fluorodeoxyglucose-avid lesions throughout the skeleton (standardized uptake value
< 2), while 68 Ga-DOTATATE positron emission tomography-computed tomography showed numerous 68 Ga-DOTATATE-avid lesions in the same region (standardized uptake value > 10)
Sarcomas are one of the less common malignancies, and despite current treatments,
patients have poor outcomes and life expectancy.[33 ],[34 ]
18 F-FDG uptake in sarcomas has been shown to be reflective of tumor biology and has
a valid predictor for tumor aggressiveness and patient outcomes.[30 ],[35 ] In addition, PET has a growing role in the evaluation of intra- and intertumoral
heterogeneity.[36 ] Piperkova et al. demonstrated advantages of 18 F-FDG PET-CT for the initial staging, restaging, and evaluation of the treatment response
for bone and soft-tissue sarcomas.[31 ] PET studies fused with cross-sectional imaging, PET-CT or PET-MRI, allow for more
accurate disease localization, detection, and as a guide for biopsies.[37 ] Furthermore, 18 F-FDG PET-CT has been shown to better differentiate soft-tissue and osseous malignancies
from benign lesions compared to PET or CT alone.[38 ],[39 ],[40 ],[41 ]
In addition to 18 F-FDG, several novel PET radiotracers have shown promising results.18 F-Fluoroestradiol,
which targets estrogen receptors (ER) has been shown to have a high sensitivity for
the detection of ER-positive skeletal metastases and is useful for quantifyingin vivo
ER expression without the need for biopsy.[42 ] Similar results have been seen for identifying osseous metastases of thyroid malignancy
with124 I.[43 ]18 F-Fluorothymidine (FLT), a radiotracer which measures tumor proliferation, has
shown promise in imaging bone and soft-tissue sarcomas.18 F-FLT can help differentiate
between high- and low-grade sarcomas and may be useful in evaluating changes in tumor
biology over time and assessing intratumoral heterogeneity.[44 ] Furthermore, the use of dual tracer “cocktail scans” are actively being investigated.
lagaru et al. have shown increased detection of osseous metastases with combined 18 F-NaF and 18 F-FDG PET-CT compared to the modalities individually.[45 ],[46 ],[47 ],[48 ]
Radiomics and artificial intelligence
Radiomics utilizes quantifiable data from imaging modalities to provide insight into
tumor biology and heterogeneity. In the era of “-omics” this data can be combined
with genetic and other data to obtain a comprehensive understanding of a patient's
tumor biology. In addition, radiomics can aid in the diagnosis of tumor cell type,
potentially negating the need for tissue biopsy in some cases and providing a better
understanding of intratumoral heterogeneity, which is an intrinsic limitation of limited
tissue sampling.[49 ],[50 ],[51 ],[52 ],[53 ] Imaging features analyzed with radiomics have been shown to have prognostic implications
for a diversity of tumors.[54 ],[55 ],[56 ],[57 ] In patients with soft-tissue sarcomas of the extremities, Vallières et al. demonstrated
an association between extracted texture features from 18 F-FDG PET-CT and a propensity for developing lung metastases.[58 ] Radiomic MRI features have also been shown to help distinguish intermediate- and
high-grade soft-tissue sarcomas.[59 ] Associations such as these aid in risk assessment at diagnosis and may help guide
first-line therapy choices.
As this field continues to grow, and imaging databases become larger, new trends may
arise from mining these large datasets. A current major limitation to the clinical
applications of radiomics is the lack of effective autosegmentation techniques, with
the majority of current studies performed with either manual or semi-automated segmentation.
However, since machine learning techniques are becoming more sophisticated, the possibility
of seamless autosegmentation in clinical practice is becoming more realistic.[51 ],[52 ],[53 ],[54 ],[55 ] Indeed, these algorithms and programs may soon be able to rapidly synthesize the
imaging data with other clinical data points to provide even more diagnostic and prognostic
information, allowing for more personalized treatment planning.[60 ],[61 ],[62 ],[63 ],[64 ]
Conclusion
Musculoskeletal malignancies have a wide array of intra- and inter-tumoral heterogeneity.
With continued advances in molecular imaging, noninvasive methods of understanding
tumor biology show promising results. This may aid in the diagnosis, prognosis, and
treatment planning and monitoring of musculoskeletal malignancies.
Declaration of patient consent
The authors certify that they have obtained all appropriate patient consent forms.
In the form the patient(s) has/have given his/her/their consent for his/her/their
images and other clinical information to be reported in the journal. The patients
understand that their names and initials will not be published and due efforts will
be made to conceal their identity, but anonymity cannot be guaranteed.
