To perform a longitudinal analysis of changes in lymph node volume and apparent diffusion coefficient (ADC) in healthy, metastatic, and hyperplastic lymph nodes.
To perform a longitudinal analysis of changes in lymph node volume and apparent diffusion coefficient (ADC) in healthy, metastatic, and hyperplastic lymph nodes.
Three groups of four female Copenhagen rats were studied. Metastasis was induced by injecting cells with a high metastatic potential in their left hind footpad. Reactive nodes were induced by injecting Complete Freund Adjuvant (CFA). Imaging was performed at baseline and at 2, 5, 8, 11, and 14 days after tumor cell injection. Finally, lymph nodes were examined histopathologically.
The model was highly efficient in inducing lymphadenopathy: subcutaneous cell or CFA inoculation resulted in ipsilateral metastatic or reactive popliteal lymph nodes in all rats. Metastatic nodal volumes increased exponentially from 5–7 mm3 at baseline to 25 mm3 at day 14, while the control node remained 5 mm3. The hyperplastic nodes showed a rapid volume increase reaching a plateau at day 6. The ADC of metastatic nodes significantly decreased (range 13%–32%), but this decrease was also seen in reactive nodes.
Metastatic and hyperplastic lymph nodes differed in terms of enlargement patterns and ADC changes. Enlarged reactive or malignant nodes could not be differentiated based on their ADC values. J. Magn. Reson. Imaging 2011;33:1151–1159. © 2011 Wiley-Liss, Inc.
THE PRESENCE OF LYMPH NODE METASTASIS is an important prognostic factor in patients with cancer. For example, in cervical cancer patients it determines (adjuvant) treatment and survival rate (1, 2). Lymph node metastasis can be a macroscopic finding during surgical lymph node dissection and confirmed by histopathological investigation; however, in 50%–80% of the low-stage cervical cancer patients lymphatic metastases measure less than 10 mm in size and are therefore infrequently detected during surgery (3). The sentinel node biopsy is a less invasive diagnostic procedure in which a radioactive isotope and blue dye are injected in or around the tumor 24 and 1 hour, respectively, before surgery to find the first radioactive and/or blue colored draining lymph node. This node is also suspected to be the first node that the tumor will drain to. This node is then removed and thoroughly examined by the pathologist (4). However, both procedures (ie, lymph node dissection and a sentinel node biopsy) are invasive, relatively expensive, and pose a risk for peri- or postsurgical complications and morbidity (5–7). A noninvasive and accurate imaging modality would therefore be highly beneficial, both to predict survival and to spare patients an invasive diagnostic lymph node staging procedure.
Magnetic resonance imaging (MRI) is a noninvasive diagnostic technique that allows 3D morphological analysis of lymph nodes. In the literature many methods are described to diagnose lymph node metastases, eg, by means of lymph node size, shape, or margins, by finding clusters of small lymph nodes, the presence of central necrosis, gadolinium contrast enhancement, or lack of USPIO contrast attenuation (8–11). The most often used malignancy criterion is increased nodal size, a clinically practical parameter to obtain. Nodal volume measured by MRI has to our knowledge not been formally evaluated as a potential malignancy criterion for MR lymph node staging.
Contrast in diffusion-weighted imaging (DWI) is determined by the thermally induced movement of water protons (diffusion) of extracellular water protons, transport, and diffusion of water protons through the cell membranes, and diffusion of intracellular water protons and microcirculation of blood. The signal intensities on DWI correlate well with changes occurring in cancerous tissue, such as increased cellularity. In hypercellular tissues the mobility of water protons is substantially diminished, resulting in hindered diffusion (12). Diffusion can be quantified by means of the apparent diffusion coefficient (ADC) (13). Tissue ADC has been hypothesized to allow detecting lymphatic metastasis in normal-sized nodes and at the same time avoid false-positive findings in enlarged reactive nodes.
The purpose of this study was to document changes in nodal volume and ADC over time for both reactive and metastatic nodes in a straightforward and reproducible experimental model. Second, the relation between ADC and volume in normal, reactive, and metastatic nodes is described.
The experiments were performed on female Copenhagen rats, weighing 150–200 g. All experiments were approved by the local Animal Care Committee. The animals were allowed food and water ad libitum. Tumor implantation and adjuvant inoculation was performed by a medical doctor (W.K.).
The R3327 MAT LyLu prostate tumor variant has been used as an experimental model for syngeneic progression and metastasis of prostate cancer. This tumor variant arose as a highly metastatic variant from the anaplastic R3327 AT variant, which had a low rate of metastasis (14). In vivo, this tumor is characterized by a rapid and hormone-independent proliferation, anaplastic histology, and metastasis to draining lymph nodes and lungs. In vitro, this tumor cell line is cultured using RPMI-1640 medium, supplemented with 10% fetal calf serum, 100 U/mL penicillin/streptomycin, 1 mmol/l sodium pyruvate, and insulin/transferrin/selenite medium supplement at 37°C in a humidified atmosphere of CO2 (5%) and air (95%) (15, 16). From subcutaneous tumor implants in flanks of male Copenhagen rats, the inguinal lymph node was populated with proliferating carcinoma cells within a week in most animals (17).
In the present experiments a tumor cell suspension containing R3327 MAT LyLu prostate tumor cells in 0.05 mL medium was prepared and inoculated aseptically in eight Copenhagen rats. The “low-dose tumor group” consisted of four rats injected with 4.0 × 104 cells and the “high-dose tumor group” included four rats injected with 2.5 × 105 cells in their left hind footpad. Rats were anesthetized with 2%–5% isoflurane in N2O/O2 (70:30) under spontaneous respiration during this procedure. The right (contralateral) popliteal node served as a benign control lymph node (“control group”). These two groups were chosen to evaluate the influence of inoculation dose on enlargement patterns.
Complete Freund's Adjuvant (CFA) is an aqueous solution of antigen emulsified in oil and killed mycobacteria. CFA enhances antibody production and induces an inflammatory response, primarily at the inoculation site. Lymphatic drainage from inoculation sites results in reactive hyperplasia of the regional nodes. This model is frequently used to create hyperplastic regional lymph nodes (18, 19). We administered 0.05 mL CFA aseptically in four female Copenhagen rats (“hyperplastic group”) in their left hind footpad. During the procedure the animals were anesthetized as described above. The right (contralateral) popliteal node served as a benign control node (“control group”).
Anesthesia was induced by intraperitoneal injection of 5.0 mL/kg midazolam and 4.5 mL/kg of a mixture of fentanyl citrate (0.315 mg/mL) and fluanisone (10 mg/mL). A baseline MRI was made (day 0) and after imaging the MAT LyLu cells (low-dose and high-dose tumor group) or CFA (hyperplastic group) were injected in the left hind footpads of all 12 rats. Imaging was performed according to the longitudinal protocol described in Table 1. Each rat was scanned six times at different timepoints for optimal information on this longitudinal study. All scans were performed on a 3.0T MR scanner (Achieva, Philips Medical Systems, Best, the Netherlands) with an eight-channel human wrist coil. After the routine localization images had been obtained, high-resolution coronal and axial T2 weighted spin-echo (repetition time (TR) 3000 msec/echo time (TE) 80 msec; TSE factor 9; section thickness 1 mm, field of view 65 × 65 mm; matrix 324 × 252; number of averages, 2) scans were acquired. We also acquired a coronal T1-weighted (TR/TE 8.1/4.1 msec; SPAIR fat suppression SPAIR offset 220 Hz, TFE factor 42; section thickness 0.5 mm, field of view 80 × 80 mm; matrix 320 × 319; number of averages 1), and a DWI dataset (parameters: TR/TE 2469/59 msec; SPAIR fat suppression, single-shot EPI factor 41; SENSE factor 2; section thickness 1 mm, field of view 60 × 60 mm; acquisition matrix (M × P) 76 × 74; number of averages 3, b values 0 s/mm2 and 1000 s/mm2).
|Low-dose tumor group (4.0*104 tumor cells)|
|High-dose tumor group (2.5*105 tumor cells)|
|Hyperplastic group (CFA)|
All MRI images were analyzed by an experienced reviewer who had knowledge of the model and the time period after cell inoculation. Image analysis was done using in-house software written in MatLab (MathWorks, Natick, MA). The volume measurements were calculated from hand-drawn regions of interest (ROIs) around the lymph nodes on each slice on which the node was visible. Volume analysis was done on the axial T2-weighted scan, which has the highest in-plane and through-plane resolution. In case of doubt, ie, partial volume effect, the coronal T2 and coronal T1-weighted images were consulted to evaluate lymph node extent from orthogonal slices. For the ADC calculation an ROI was drawn on the b = 1000 s/mm2 DW image within each lymph node at the slice in which the largest ROI could be drawn. This ROI was copied to the ADC map. The ADC map was calculated by fitting the signal of the diffusion weighted images to the Stejskal–Tanner Equation (20): S(b) = S0 * e −b*ADC to the data points at 2 b-values: 0 and 1000 mm2/s, with S0 and ADC as fit parameters. In Figures 1 and 2 an example of the methods used for the volume and ADC measurements, respectively, are presented.
Further analysis consisted of correlating the ADC to the volume of metastatic, hyperplastic, and normal lymph nodes. Because ADC depends on the tissue cellularity, cell membrane permeability, and other factors such as extracellular matrix composition (12, 13), ADCs may differ per volume and per node type (metastatic, hyperplastic, and normal). This was measured in all lymph nodes at all timepoints.
Rats were sacrificed with a lethal dose of sodium pentobarbital 14 days after tumor cell or adjuvant inoculation or prior to day 14 if metastatic nodes reached a volume of >25 mm3. Sodium pentobarbital was administered intraperitoneally while the animals were still under anesthesia. The popliteal lymph nodes were identified and dissected along with the surrounding tissue, labeled for location (left or right popliteal node), and fixed with 10% formalin. The lymph nodes were subsequently embedded in paraffin and the sections were stained with hematoxylin and eosin for microscopic examination by an experienced pathologist. The pathologist noted the presence of tumor cells, necrosis, and reactive hyperplasia. In metastatic nodes, the percentage of lymph node involved by tumor cells was noted as well.
The measurements of the contralateral control lymph nodes (“control group”) of rats in the low-dose tumor group, high-dose tumor group, and hyperplastic group were combined because no statistical significant differences in volumes or ADCs were found (analysis of variance [ANOVA], P > 0.05). The volume and ADC data are described using summary statistics and mean values are presented. The differences between two groups were tested by t-tests, while the three (low-dose tumor group, high-dose tumor group, and control lymph nodes) or four groups (low-dose tumor group, high-dose tumor group, hyperplastic group, and control group) were tested using ANOVA statistics. Correlation of ADC and volume was tested by linear regression analysis. P < 0.05 was considered statistically significant. Data were analyzed using the Statistical Package for the Social Sciences v. 15.0 (SPSS, Chicago, IL).
Histopathological analysis of the 24 popliteal lymph nodes in 12 female Copenhagen rats confirmed ipsilateral metastatic nodes in all eight tumor-injected rats, hyperplastic nodes in all four CFA-injected rats, and normal nodes in the contralateral leg of the 12 animals. Focal central necrosis was found in 2/8 metastatic and 0/4 hyperplastic nodes. In the hyperplastic nodes, macrophages with phagocytosed particles, eosinophils, neutrophils, sinus histiocytosis, and in one lymph node granulomas were found. All metastatic nodes showed patterns of secondary follicle formation. Figure 3 presents the pathological findings.
As shown in Figs. 4 and 5 and in Table 2, enlargement of the metastatic lymph nodes was detectable 2 days after tumor cell inoculation (P = 0.050). In the three categories of lymph nodes (ipsilateral nodes in the low-dose tumor group vs. ipsilateral nodes in the high-dose tumor group vs. contralateral control nodes) the lymph node volume became significantly different from each other 14 days after cell inoculation (Fig. 4). Enlargement of metastatic lymph nodes showed an exponential pattern. The lymph nodes in the high-dose tumor group enlarged faster than nodes in the low-dose tumor group; eventually in both groups the lymph nodes reached the same size (25 mm3), although at a different (later) timepoint (data not shown). Hyperplastic nodes rapidly increased in size and lymph node volumes differed significantly from the control nodes within 24 hours after CFA administration. Six days after CFA inoculation the lymph node volume reached a plateau at ≈40 mm3 (Fig. 5). The enlargement patterns of metastatic lymph nodes appeared to be different from that of reactive nodes, showing continued exponential enlargement during the observation period. The contralateral lymph nodes remained at their baseline size: 5–7 mm3.
|Volume (mm3)||Baseline||Day 1||Day 2||Day 4/5||Day 8||Day 11||Day 14|
|Low-dose tumor group||4.9||7.3||8.8||10.1||11.8||14.5|
|High-dose tumor group||7.2||7.4||9.4||14.3||18.9||25.7|
|ADC × 10−3||Baseline||Day 1||Day 2||Day 4/5||Day 8||Day 11||Day 14|
|Low-dose tumor group||0.69||0.69||0.65||0.60||0.51||0.59|
|High-dose tumor group||0.79||0.65||0.64||0.54||0.49||0.52|
Metastatic as well as reactive lymph nodes showed a decrease in ADC after cell inoculation (Table 2, Fig. 6). As the mean ADC of control lymph nodes varied between 0.70 × 10−3 and 0.80 × 10−3 mm2/s, the ADC of metastatic nodes dropped to 0.50 × 10−3–0.60 × 10−3 mm2/s within 8 days (P = 0.005 for metastatic vs. control nodes, t-test). The ADC of hyperplastic nodes also dropped to 0.50 × 10−3–0.60 × 10−3 mm2/s and this decrease was already seen at 4 days (P = 0.023 for hyperplastic vs. control nodes, t-test). The percentage decrease in ADC was 13%–32% ((ADC day 8/Baseline ADC) × 100%) for metastatic nodes at day 8 after injection, versus a decrease of ≈24% ((ADC day 8/Baseline ADC) × 100%) in the reactive nodes in the same time period. The ADC decrease in the metastatic lymph nodes remained stable between day 8 and 14, while the ADC decrease in hyperplastic nodes tended to drop further with increased nodal volume.
The correlation (r2) of ADC to volume of lymph nodes was quite similar for metastatic and hyperplastic lymph nodes (Fig. 7): 0.38 versus 0.46, respectively. This means that 38% of the metastatic node's ADC value can be explained by its volume versus 46% for reactive nodes.
Injection of CFA or tumor cell suspension constitutes an efficient experimental model to study normal, hyperplastic reactive, and metastatic nodes. Subcutaneous cell or adjuvant inoculation resulted in ipsilateral metastatic or reactive popliteal lymph nodes in all experimental animals. This model was previously used successfully by Vassallo et al (19) to study the uptake of ultrasmall iron particles in lymph nodes. The R3327-MATLyLu variant of the Dunning rat prostate carcinoma showed rapid growth and almost invariably metastasized to lymph nodes within 10–15 days after subcutaneous implantation (14, 19). Another well-known model to create lymphatic metastasis employs New Zealand white rabbits and VX2 squamous cell carcinoma cells to resemble head and neck cancer (21). The use of R3327 MATLyLu cells has several advantages over VX2 carcinoma cells: smaller animals (Copenhagen rats) are better characterized, less expensive, easier to handle, and are therefore more practical than larger species. Also, in the footpad model there is only a single popliteal lymph node in the anatomic structure of the murine rear leg and the anatomic location of the popliteal lymph node is constant, superficial, and easy to identify by palpation and MRI. In our model, all measurements were done using a clinical MR scanner with sequences available in daily clinical practice which aids in comparing the results to the clinical setting.
When using size criteria in nodal staging, false-positive results may be caused by lymph nodes that are swollen due to nodal inflammation induced by upstream tumor presence or as a reaction to prior diagnostic procedures like tumor biopsy. The term for this condition is tumor-reactive lymphadenopathy (22). In addition, lymphadenitis may result from inflammation unrelated to tumor presence. Vassallo et al (19) also used CFA to induce lymphadenitis in draining lymph nodes and compared the results of metastatic and reactive nodes in lymph node MR staging. Comparing metastatic lymph nodes with hyperplastic lymphadenopathy was not performed in other experimental studies evaluating lymph node metastases (23). While grossly enlarged lymph nodes pose the fewest problems, differentiating small metastatic from small hyperplastic and control lymph nodes is of crucial importance in lymph node staging in cancer patients. As described earlier, CFA is an aqueous solution of antigen emulsified in oil and killed mycobacteria. It has been used over the past 45 years to stimulate polyclonal antibody production in animals. In addition, it has been shown to consistently induce local granulomatous reactions and lymphatic drainage from inoculation sites, resulting in reactive hyperplasia of the regional nodes. However, it was not described previously in the literature that reactive lymphadenitis induced by CFA administration resembles reactive lymphadenitis as a response to tumor-associated proteins.
The volume of metastatic lymph nodes increased exponentially depending on the number of cells administered: when 2.5 × 105 tumor cells were injected the node reached a volume of 25 mm3 in 14 days, and with 4.0 × 104 injected cells the volume was 14 mm3 in 14 days. The contralateral control lymph node remained at baseline volume: 5 mm3. The reactive lymph nodes grew in a different pattern: a fast increase in volume within 24 hours after CFA inoculation, which reached a plateau 6 days after injection at 40 mm3.
Total lymph node volume measured by computed tomography (CT) has been studied for correlation with treatment response and differentiating benign from malignant disease in head and neck cancer patients (24, 25). Liang et al (25) found that tumor volume measured by CT in patients with head and neck cancer was higher in patients with cervical metastases than those without cervical metastases; however, it was not an independent factor associated with metastases after controlling for other variables like central nodal necrosis and short axis diameter > 7.4 mm. Murray et al (26) measured an oblique sagittal nodal area (cm2) on MRI scans and found a sensitivity and specificity of 100% and 56%, respectively, in predicting lymph node metastasis when area measurement was used in combination with gadolinium enhancement. Unlike tumor volume measurements for prognostic purposes, to our knowledge no literature could be found in which lymph node volume was directly evaluated as a malignancy criterion for lymphatic metastasis.
The ADC of metastatic lymph nodes significantly decreased (range 13%–32%) 8 days after cell inoculation, while a significant decrease (24%) in ADC in hyperplastic reactive nodes was seen within 4 days after injection. After the initial decrease, ADC values of both metastatic and reactive nodes remained at that level 1 week after injection.
ADC of transplanted squamous cell tumors has been studied before in a mouse model (12). They found a decrease in ADC during tumor growth which tended to increase again when tumor necrosis developed. These results in primary squamous cell tumors are comparable to our results of ADC in metastatic lymph nodes; we did not sacrifice rats at all timepoints, but at day 14 necrosis was present in only 2/24 nodes.
ADC values were just partly related to nodal volume, as evidenced by the low Pearson correlation coefficients: only 38% of the ADC value in metastatic nodes could be explained by their volume. For hyperplastic reactive nodes this percentage was somewhat higher: 46%. There was no significant difference in ADC between hyperplastic and metastatic nodes. Because ADC depends on the tissue cellularity, cell membrane permeability, and other factors such as extracellular matrix composition (12, 13), our model suggests that hyperplastic and metastatic nodes seem to resemble each other in terms of these combined properties. The histopathological results point in this direction as well. ADC measurement was not sufficient in differentiating small metastatic from small hyperplastic or control lymph nodes. This is relevant when studying lymph node ADC in patients with possible lymph node metastasis.
A prerequisite for adequate diagnostic accuracy is reproducibility. Fabel et al (27) studied reproducibility of CT-measured nodal volume in patients with melanoma. They compared lymph node diameters measured by response evaluation criteria in solid tumors (RECIST) and semiautomated nodal volumetric analysis in terms of observer variability; the RECIST measured differences between two observers were ≈0.9 cm; for the volumetric analyses this was very much improved to 0.2 cm. Therefore, it seems that volume measurements allow for a higher precision than axis diameter measurements. Reproducibility of ADC measurements varies in the literature. While Koh and Collins (13) found excellent reproducibility of ADC measured in solid tumors, ADC reproducibility of normal human lymph nodes was not optimal; Kwee et al (28) found significant heterogeneity. We did not perform a reproducibility study; the purpose of this study was to investigate the relation of reactive and metastatic lymphadenopathy to volume and ADC measurements. Although reproducibility was not a subject of our study, it should be kept in mind in interpreting the results.
Even though longitudinal analysis allows for comparison with baseline values, a limitation of our study is the relatively small number of nodes studied. Another limitation of our study was that the ROIs were drawn on the whole lymph nodes and no necrotic tissue was excluded from metastatic or reactive nodes. While this may have affected ADC values, it should be noted that necrosis was only seen in 2/24 examined lymph nodes and the percentage necrotic tissue was limited in these nodes.
In conclusion, metastatic and hyperplastic lymph nodes differed in terms of enlargement patterns and ADC changes. Volume was a better predictor of lymphatic metastasis than ADC and the results suggest that nodal volume increases before ADC decreases in metastatic lymph nodes. Enlarged reactive nodes could not be differentiated from enlarged malignant nodes based on ADC values. These findings may be relevant when interpreting MRI examinations of patients with possible lymph node metastasis.