Comparison of abdominal ultrasound and magnetic resonance imaging for detection of abdominal lymphadenopathy in dogs with metastatic apocrine gland adenocarcinoma of the anal sac


  • Presented in poster format at the Veterinary Cancer Society Annual Conference 2010 in San Diego, CA – Winner of the Resident Clinical Science Award


Imaging studies in humans with anal and rectal cancer indicate that magnetic resonance imaging (MRI) is a more sensitive technique than abdominal ultrasound (AUS) for the detection of abdominal lymphadenopathy. The purpose of this retrospective study was to directly compare the efficacy of these two techniques in detecting abdominal lymphadenopathy in dogs with apocrine gland adenocarcinoma of the anal sac (AGAAS). Six dogs with histologically confirmed AGAAS and histopathologic confirmation of metastasis to abdominal lymph nodes (LNs) had AUS and abdominal MRI. AUS identified lymphadenopathy in two of six dogs, whereas MRI identified lymphadenopathy in all the six dogs. Lymphadenopathy was predominantly sacral in location, with involvement of the medial iliac and hypogastric LNs in only two cases. These data suggest that MRI is more sensitive than AUS for detecting sacral abdominal lymphadenopathy in dogs with AGAAS. As such, MRI could be considered in any patient with AGAAS for initial staging of this disease.


Canine apocrine gland adenocarcinoma of the anal sac (AGAAS) is a relatively uncommon tumour, accounting for approximately 17% of all perianal tumours and 2% of all skin tumours in the dog.[1-3] Apocrine gland adenocarcinoma, which arises from the apocrine glands in the walls of the anal sacs, is typically a disease of older dogs, with the mean age at the time of presentation reported to be 10–11 years with equal sex distribution. There is limited information on breed predisposition, although the German shepherd, English cocker spaniel, dachshund, Alaskan malamute and English springer spaniel may be at increased risk of developing AGAAS.[1-3]

Apocrine gland adenocarcinomas in the dog are described as both invasive and metastatic in nature, with metastatic rates reported to range from 36 to 96% at the time of presentation, with regional lymph nodes (LNs) being the most common site.[3] Metastasis to the lung, liver, spleen, bone and less commonly to the mediastinum, heart, pancreas, adrenal glands or kidneys may also occur.[1-4] Additional morbidity is associated with hypercalcaemia of malignancy in 25–53% of the cases, which is mediated by tumour secretion of parathyroid hormone-related peptide.[1] Clinical signs of AGAAS may vary and are commonly related to the size of the tumour at initial diagnosis, the presence of hypercalcaemia and the presence of sublumbar nodal metastasis causing obstruction of the pelvic canal.[1-4]

Staging of AGAAS typically consists of a complete blood count, chemistry panel and urinalysis, as well as thoracic radiographs to evaluate for evidence of pulmonary metastatic disease or mediastinal involvement. Abdominal ultrasound (AUS) is routinely used to evaluate the size of regional LNs and the echogenicity of other abdominal organs, especially the spleen and liver. Multiple treatment options have been described for the treatment of AGAAS, consisting of surgery of both primary and secondary tumours as well as regional LNs; platinum drug chemotherapy; mitoxantrone drug chemotherapy; doxorubicin drug chemotherapy; melphalan drug chemotherapy; targeted therapy including toceranib phosphate (Palladia®) and radiotherapy of both primary tumours and regional LNs.[1-5] A study of 113 dogs with AGAAS in varying stages of disease treated with differing treatment modalities reported a median overall survival time of 544 days for all treated dogs.[3]

LN status is important for therapeutic planning, and may be an important prognostic factor regarding long-term survival, although studies are lacking supporting a poorer prognosis in dogs with AGAAS and metastasis to the LNs versus those without LN metastasis. Accurate pretreatment knowledge of LN status may be helpful in determining prognosis and planning the optimal extent of lymphadenectomy. In addition, pretreatment knowledge of LN status would help in selecting patients who would benefit from adjuvant chemotherapy and radiation therapy. The LNs responsible for drainage of the pelvic and perineal region are those of the iliosacral lymphocentre. It is composed of the medial iliac, hypogastric and sacral LNs. The medial iliac LNs are the largest of the group, lying between the caudal circumflex and external iliac arteries, adjacent to the aorta and vena cava, ventral to the fifth or sixth lumbar vertebral bodies. The hypogastric LNs are smaller, often paired and lie within the angle between the internal iliac and median sacral arteries, ventral to the sixth or seventh lumbar vertebral bodies. The sacral LNs are the most caudally located and lie ventral to the sacrum, alongside the median sacral artery.[6]

AUS is often the test employed to screen for abdominal lymphadenopathy in dogs with AGAAS.[1-3, 7] The ultrasonographic features of normal and abnormal medial iliac LNs have been described, including the appearance of LNs specifically affected by metastasis of AGAAS.[8-12] Normal medial iliac LN size is directly proportional to body size, with mean widths ranging from 0.43 cm for small breeds up to 0.89 cm for larger breeds.[9] Normal LNs have a fusiform shape, homogeneous echotexture and an echogenicity that is homogeneous, and isoechoic to slightly hypoechoic to the surrounding fat.[9] Abnormal LNs affected by both benign and malignant disease are larger and more numerous than normal LNs, and demonstrate a variety of altered imaging characteristics, including hypoechogenicity, heterogeneity, rounded shape and distorted hilar fat appearance.[10-12] Rounded shape and heterogeneity may help to distinguish benign from malignant transformation.[11, 12] Specifically, LNs affected by AGAAS metastasis are described as being larger and more numerous, with a rounded shape and a reduced and heterogeneous echogenicity.[10] Considerable overlap exists in the ultrasonographic appearance of LNs affected by benign and malignant disease, and cytology or histology is necessary for definitive diagnosis.[12] Cytologic evaluation of medial iliac LNs affected by AGAAS is often possible with ultrasound guidance, although sampling of the hypogastric LNs can be more difficult because of their deeper and more caudal location and close proximity to the adjacent vasculature.

In human medicine, magnetic resonance imaging (MRI) and computed tomography (CT) have been used as a standard screening test for anal and rectal cancer and have been found to be more sensitive than AUS for detecting abdominal lymphadenopathy. Morphological characteristics such as a round shape, irregular borders and heterogeneous signal intensity, as well as LNs of 1 cm or over, are suggestive of nodal involvement in human MRI studies.[13, 14] Descriptions of the MRI characteristics of normal and abnormal abdominal LNs in dogs are lacking in the veterinary literature. Normal canine head and neck LNs have been described as being homogeneous and hypointense to surrounding fat on precontrast T1-weighted sequences, isointense on postcontrast T1-weighted sequences, slightly hypointense on T2-weighted sequences and hyperintense to surrounding fat on short-tau inversion recovery (STIR) sequences.[15] Similar to ultrasound, LNs are often considered abnormal on MRI if they are increased in size or number, have a rounded shape, or if they have an altered or heterogeneous signal intensity or abnormal contrast enhancement pattern. Although a specific reference describing the MR characteristics of either normal or abnormal iliosacral LNs was not found during a search of the veterinary literature, LNs are often considered abnormal on MRI if they are increased in size or number, have a rounded shape, altered or heterogeneous signal intensity, or if they demonstrate heterogeneous contrast enhancement pattern.

Systematic studies comparing AUS and MRI techniques in veterinary patients with AGAAS have, to our knowledge, not been reported. As imaging technology continues to evolve, the purpose of this study was to systematically review the current role of assessing LN status in canine AGAAS patients using AUS and abdominal MRI. We hypothesized that MRI would be a more sensitive imaging modality for identifying lymphadenopathy in dogs with AGAAS, specifically pelvic lymphadenopathy. This retrospective study reviews the role of these two imaging modalities in determining those patients with lymphadenopathy.

Materials and methods

Between January 1999 and August 2009, 88 canine AGAAS patients presented to the Veterinary Teaching Hospital at Washington State University. They were included in this study if they had both an AUS and abdominal MRI exam as part of their pretreatment diagnostics and a histopathologic diagnosis of AGAAS and histopathologic confirmation of metastasis to the abdominal LNs. Six dogs fulfilled all criteria to be entered into this investigation.

All ultrasound examinations were performed by an American College of Veterinary Radiology (ACVR) board-certified radiologist or by a radiology resident under supervision of an ACVR diplomate, with the specific goal of evaluating for lymphadenopathy at the time of initial presentation. Five of the ultrasound exams were performed at Washington State University using a single ultrasound machine equipped with multifrequency curvilinear (4–9 MHz) and linear array (5–13 MHz) transducers (MyLab 70; Esaote North America, Indianapolis, IN, USA). The remaining exam was performed by an ACVR diplomate at a referral clinic. Criteria used for ultrasonographic abdominal lymphadenopathy included enlargement, rounded shape and reduced or heterogeneous echogenicity, as has been previously described.[10-12] LNs were considered enlarged if they measured greater than 1 cm for medial iliac nodes and 7 mm for hypogastric nodes. Because of the small number of patients included in this study and lack of standardization of images, no attempt was made to include additional criteria for determining normal from abnormal nodes.

All MRI exams were performed at Washington State University and were interpreted by an ACVR diplomate. All MRI studies were performed on the same 1.0-T magnet (Phillips Gyroscan 1.0 T; Phillips Healthcare, Andover, MA, USA) using an appropriate surface coil matched to the size of patient. STIR sequences of the caudal abdomen and pelvic region obtained in transverse, sagittal and dorsal planes were acquired in all studies. T1 sequences acquired before and after administration of gadolinium-based IV contrast (gadopentetate dimeglumine, Magnevist; Bayer Healthcare Pharmaceuticals, Wayne, NJ, USA) were also included in all exams, with all three planes being acquired postcontrast. Slice thickness ranged from 4 to 9 mm. Additional sequences were included in some exams at the request of the attending radiologist, but as this was inconsistent, only T1 and STIR sequences were reviewed for the purposes of this study. Criteria for MRI abdominal lymphadenopathy included enlargement (greater than 1 cm diameter for medial iliac nodes and greater than 7 mm diameter for hypogastric and sacral LNs), rounded shape, heterogeneous signal intensity and heterogeneous contrast enhancement.

In all cases, AUS preceded MRI, with a maximum of 21 days between studies (median 3 days, range 1–21 days). The imaging studies were reported at the time of investigation, with both ultrasonographic and MRI features being described in detail by board-certified veterinary radiologists. Following compilation of the study cases, an ACVR board-certified veterinary radiologist (G. D. R.) and a third-year veterinary radiology resident (C. S. M.) reviewed all available imaging reports and images, and by consensus review, determined the presence or absence of lymphadenopathy. Reports and images were available for review for all MRI exams and five of the ultrasound exams. A detailed report, including detailed description and measurements of the iliosacral LNs, was available for the single ultrasound exam performed away from Washington State University.



Patients included were two castrated males and four spayed females of five different breeds, including Labrador Retriever, English Cocker Spaniel, German Shorthaired Pointer, Great Pyrenees and Tibetan Terrier. The median age was 7 years (range 7–12) and the median body weight was 35 kg (range 14–38).

Abdominal MRI

Abdominal MRI detected lymphadenopathy in all the six dogs included in this study. Histopathologic evaluation of these LNs following surgical excision confirmed AGAAS metastasis in all six of these patients. The sacral LNs were most commonly involved, being considered abnormal in all the six cases where lymphadenopathy was detected on MRI. The hypogastric LNs were considered abnormal in three cases, whereas the medial iliac LNs were abnormal in only two cases. Sole involvement of either the medial iliac or hypogastric LNs was not identified in any patient. All LNs classified as abnormal were enlarged. LNs ranged in thickness between 2.0 and 3.8 cm for medial iliac nodes, 1.6 and 2.3 cm for hypogastric nodes and 1.4 and 6.5 cm for sacral nodes. In addition to enlargement, consistent findings involving abnormal LNs on MRI exam included heterogeneous signal intensity on STIR sequences, a rounded shape, irregular or undulating margination and a heterogeneous contrast enhancement pattern. Asymmetry between left and right sides was also often identified, with the nodes on the side ipsilateral to the primary tumour being more affected. Local invasion into adjacent intrapelvic soft tissue structures was suspected in several cases, and marked narrowing of the pelvic canal and colonic displacement and compression was often present secondary to severe sacral lymphadenopathy.

Abdominal ultrasound

Abnormal iliosacral LNs were identified in only two of six ultrasound examinations. In both of these cases, medial iliac and hypogastric involvement was present, as shown in Fig. 1. In two additional cases where there was no evidence of either medial iliac or hypogastric lymphadenopathy on ultrasound exam, the cranial aspect of a large and heterogeneous mass extending into the pelvic canal was identified. Review of corresponding MRI exam in these two cases allowed classification of this mass as a grossly enlarged sacral LN, although lack of vascular or other anatomic landmarks prevented classification as such by the attending radiologist at the time of initial ultrasound exam. This is illustrated in Figs 2 and 3. All LNs classified as abnormal were enlarged when compared with reported normal values, with the smallest measuring 1.7 cm in thickness. All abnormal LNs were rounded in shape, had decreased and/or heterogeneous echogenicity and asymmetry was noted from right to left, with the side ipsilateral to the primary tumour being more affected. Most abnormal LNs also had an undulating margin. Evidence of local osseous invasion was not identified in any of the cases.

Figure 1.

Ultrasound (A) and MR (B) images obtained from a dog with right-sided medial iliac lymphadenopathy identified on both ultrasound and MRI. Hypogastric lymphadenopathy was identified on both ultrasound and MRI, whereas sacral lymphadenopathy was identified in this patient only on MRI. (A) Dorsal plane ultrasound image of the medial iliac LN located just lateral to the bifurcation of the aorta and right external iliac artery. The LN is enlarged, rounded in shape, heterogeneous and hypoechoic in echogenicity (white arrow). Cranial is to the left. (B) Dorsal plane T1-weighted postgadolinium image of the medial iliac LN located just lateral to the right external iliac artery. The LN is enlarged, rounded in shape, has mildly undulating margination and is heterogeneously contrast enhancing (black arrow). Right is to the left side of image, cranial is to the top. (TR 500.0 ms, TE 18.0 ms, slice thickness 6.0 mm, interslice gap 1.0 mm, FOV 25.0 cm × 25.0 cm and matrix 256 × 256).

Figure 2.

Sagittal plane T1-weighted postgadolinium MR images of the lumbosacral region of two dogs with sacral lymphadenopathy and normal medial iliac and hypogastric LNs. (A) This sacral LN (black arrow) was identified only with MRI. The LN is contrast enhancing caudally, with cavitation cranially. The contrast-enhancing left-sided anal sac mass is identified to the right of the image (white arrowhead). Cranial is to the left of the image, dorsal is to the top. (TR 539.3 ms, TE 15.0 ms, slice thickness 6 mm, interslice gap 1 mm, FOV 25.9 cm × 25.9 cm and matrix 256 × 256). (B) This sacral LN (white arrow) was identified on ultrasound as a poorly defined mass protruding cranial to the pelvic inlet. The LN is fairly homogeneous in contrast enhancement with well-defined margination. Cranial is to the left of the image, dorsal to the top. (TR 500.0 ms, TE 18.0 ms, slice thickness 6 mm, interslice gap 2 mm, FOV 24.9 cm × 24.9 cm and matrix 256 × 256).

Figure 3.

Ultrasound (A) and MR (B and C) images obtained from a dog with sacral lymphadenopathy identified only on MRI. Medial iliac and hypogastric LNs were within normal limits on ultrasound and MRI. (A) Ultrasound of normal left medial iliac LN is outlined by cursors. Cranial is to the left, ventral is to the top. (B) Dorsal plane T1-weighted postgadolinium image of aortic trifurcation demonstrating normal left (black arrow) and right medial iliac LNs. Right is to the left side of image, cranial is to the top. (TR 400.0 ms, TE 17.0 ms, slice thickness 6.0 mm, interslice gap 1.0 mm, FOV 29.9 cm × 29.9 cm and matrix 512 × 512). (C) Transverse plane T1-weighted postgadolinium image at level of cranial acetabula demonstrating markedly enlarged left sacral LN (white arrow), with milder enlargement of the right sacral LN. The LNs have thick and irregularly marginated contrast-enhancing capsules with non-enhancing centres. The colon (thin black arrow) is being displaced rightwards and ventrally and the lateral aspect of the left LN is effacing the margins of the adjacent musculature. Right is to the left side of the image, dorsal is to the top. (TR 942.9 ms, TE 14.0 ms, slice thickness 4 mm, interslice gap 1 mm, FOV 22.9 cm × 22.9 cm and matrix 256 × 256).


Preoperative imaging is necessary in dogs with AGAAS to select candidates for surgery, to plan surgery, to plan adjuvant radiation therapy and to be able to inform owners of the extent of surgery and associated prognosis. All dogs with lymphadenopathy detected on ultrasound were also detected with MRI. This finding indicates that AUS failed to detect 67% of abdominal lymphadenopathy detected on abdominal MRI. In total, eight abnormal iliosacral LNs were detected on MRI that were not identified on ultrasound (six sacral and two hypogastric).

In this report, the sacral LNs were considered to be affected by metastasis in more dogs than were either the hypogastric or medial iliac LNs. The sacral LNs are anatomically nearest to the anal glands, and we hypothesize that they may therefore represent the initial site of metastasis for the majority of dogs with AGAAS, with the hypogastric and medial iliac LNs being affected later on in the course of the disease. The afferent lymph vessels of the sacral LNs come from adjacent musculature and viscera. Efferent vessels then travel as a plexus to the hypogastric LNs.[6] We are unaware of a reference describing a predilection for sacral LN metastasis in AGAAS patients, and more studies involving greater numbers of patients would be necessary for confirmation of this trend identified in these six dogs.

Normal sacral LNs are typically not amenable to imaging with AUS because of their intrapelvic location immediately ventral to the sacral body. If severely enlarged, as was the case in two of the patients in this report, the cranial-most aspect of the nodes may be identified on ultrasound exam extending cranial from the pelvic inlet. Less severely enlarged LNs, however, are unlikely to be identified. In both of these cases, lack of vascular or other anatomic landmarks prevented classification of these masses as enlarged sacral LNs on AUS exam, whereas they were clearly identified as such on MRI exam. The possibility of these structures representing cranial extension of the primary mass could not be excluded based on ultrasound findings, and therefore these two patients were not considered to have evidence of lymphadenopathy on ultrasound exam.

In comparison to the sacral LNs, the medial iliac and hypogastric LNs are accessible to the ultrasonographer for evaluation.[8-10] The medial iliac LNs, being the largest and most cranial of the group, can be routinely visualized in normal dogs, and the normal appearance of these nodes has been described.[8, 9] The hypogastric LNs are smaller, more caudally positioned and are described as being less frequently detected on routine exam,[8] although studies describing the specific appearance, size and rate of detection of these nodes are lacking. The characteristics of abnormal medial iliac LNs, including those affected by metastasis from AGAAS, have also been described in detail, with numerous criteria being suggestive of malignancy.[10-12] Similar reports concerning the hypogastric LNs are not available in the veterinary literature. In our experience, abnormal hypogastric LNs are most often detected in conjunction with abnormal medial iliac LNs. The findings of this report indicate that the medial iliac and hypogastric LNs are the least likely of the iliosacral group to be affected by metastasis from AGAAS. Although medial iliac LNs affected by AGAAS are likely to be detected by ultrasonography, the finding of normal-appearing medial iliac LNs clearly does not rule out the potential for metastasis of AGAAS to the other LNs of the iliosacral centre. As such, it is imperative to always critically evaluate the region of the hypogastric LNs even when the medial iliac LNs are considered ultrasonographically normal.

In comparison to ultrasound, the cross-sectional and multiplanar nature of image acquisition and superior contrast resolution of MRI ameliorate anatomical and operator limitations that are seen with ultrasound, making MRI ideal for the imaging of the iliosacral LNs. As demonstrated in this report, MRI is superior for evaluation of all LNs of the iliosacral centre, especially the sacral LNs, the most common of the group to be involved. Additionally, MRI exam allows for simultaneous imaging of the primary tumour, providing invaluable information to help direct surgical intervention or radiation planning. The major disadvantages of MRI include the added cost and necessity for general anaesthesia. CT may be a viable option in lieu of MRI to image the iliosacral LNs, the liver and spleen and other abdominal organs, as well as the primary tumour, at a lower cost than most MRIs and associated with shorter anaesthesia times. Although the CT appearance of presumed normal abdominal LNs has been described, further investigation is necessary to determine the utility of CT in the diagnosis of lymphadenopathy in AGAAS patients.[16]

MRI has previously been shown to be more sensitive than AUS for better characterization of the primary tumour and detection of lymphadenopathy in people.[13, 14, 17] Specifically, one study evaluating current trends in staging rectal cancer determined MRI to be the best overall imaging modality for local staging, with ultrasound, specifically endorectal ultrasound, a less favoured modality.[13] Endorectal ultrasound has not been described in dogs and the potential utility is not known. The modality could be explored for staging the pelvic LNs, however may not be very practical, as it would require preparation in the form of enemas, and heavy sedation or anaesthesia to perform given that these dogs are painful and have narrowed pelvic canals because of LN size and the presence of perirectal masses. Evaluation of abdominal tumours and subsequent staging of these tumours in children also rely heavily on the use of MRI versus ultrasound.[17] To our knowledge, this is the first report on comparing these two imaging modalities for the identification of abdominal lymphadenopathy in veterinary medicine.

Although all patients in this study had histological confirmation of AGAAS and histological confirmation of metastasis to abdominal LNs, in practice, we often have to make clinical decisions without histologic or cytologic confirmation based on imaging findings. In humans with anal cancer, which is rarely operative, the presence of lymphadenopathy on MRI is considered evidence of nodal involvement. An investigation of six operated patients revealed understaging in only one patient.[14] Given that there are currently no MRI or ultrasound findings known to be pathognomonic for LN metastasis, LN sampling, either by fine needle aspiration or by biopsy, is necessary for confirmation of metastasis in all cases.

There are some inherent limitations associated with our inclusion criteria for determining the presence of lymphadenopathy in this study. One limitation is the lack of validation of our criteria used for detection of lymphadenopathy on MRI. There is a lack of published data on the MR appearance of iliosacral LNs and lack of size references for hypogastric and sacral LNs on both ultrasound and MRI. Our inclusion criteria for MRI lymphadenopathy are based on extrapolation from the limited reports of MR LN characteristics that are available,[15] as well as from personal experience. We extrapolated the 1 cm size for medial iliac nodes from ultrasound data. We chose a cutoff of 7 mm as normal for hypogastric and sacral LNs based on the reports describing these nodes as consistently smaller than the medial iliac LNs.[16] Additionally, despite reported data demonstrating positive correlation of normal LN size, as measured by ultrasound, varying directly with body weight, patient size was not considered when determining cutoff values in this study. We believe this is reasonable based on the description of malignant LNs in the literature being consistently larger than 1 cm in a variety of breeds.

A potential limitation to this study includes bias when assessing LNs both on AUS and abdominal MRI, as the radiologists knew the patients had a diagnosis of AGAAS prior to reviewing the studies. In addition, a potential limitation exists in the direct comparison of AUS and abdominal MRI in detecting sacral lymphadenopathy. As we have previously stated, AUS may not identify sacral LNs based on the acoustic principles of the pelvis, whereas MRI would not be affected by this anatomic limitation. In the cases where the medial iliac LNs were affected, MRI and ultrasound both were able to detect the imaging abnormalities that categorized these nodes as being abnormal. We can therefore only conclude from the findings of this study that MRI is superior because of its ability to image the sacral LNs and not because of a greater sensitivity in detecting subtle changes in LN imaging characteristics when compared with ultrasound.

The findings of this study indicate that abdominal MRI appears to be a more sensitive technique than AUS for detecting lymphadenopathy in dogs with AGAAS. In this study, iliosacral lymphadenopathy was detected in the majority of dogs diagnosed with AGAAS, with the sacral LNs being most commonly involved. Given that sacral lymphadenopathy is unlikely to be detected with AUS, MRI should be considered in any patient with AGAAS in conjunction with LN cytology or biopsy to more reliably stage the disease, particularly when accurate characterization of the extent and distribution of metastasis affects therapeutic planning.