In basic comparative studies (Mohr and Weiûleder 1996), lymphotropic iron oxide compounds with storage in sinus histiocytes oflymph nodes and gad- olinium (iD) DTPA-PGM were tested; the results in- dicated that both categories ofcompounds improved the differential diagnosis between tumor-free and cancer-infiltrated lymph nodes, irrespective of size.
While iron oxides are already in uses in clinical studies, GDDTPA-PGM is still being evaluated in preclinical study programs.
After various experimental studies on different sites ofprimaries, a first stringently statistically controlled study has been now performed to evalu- ate USPIO in the diagnosis in breast cancer pa- tients (Stets et al. 2000, 2002). Qualitative and quantitative parameters were evaluated in axillary lymph nodes by USPIO contrast agent and MR.
The studies were planned on the basis ofexperi- ence in different lymph node basins, referred to other tumor categories.
Data from two scientific centers taking part in a multicenter trial were obtained (Heywang-Kæbrun- ner et al. 2002). Each ofthe diagnostic centers used the same techniques. This made it possible to pool the cases.
It was shown that cancer-infiltrated node areas with destruction ofthe pre-existent node architec- ture were characterized by markedly less iron ox- ide storage than elsewhere, caused by the loss of the so-called sinus histiocytes that are active in
phagocytosis. According to experiences already available, the presence ofUSPIO in preserved lymph node structures can be detected with opti- mal accuracy by MRI. Iron oxide acts as a para- magnetic agent and causes signal drop out, partic- ularly in T2-weighted and T2-sequences.
In the two radiodiagnostic institutes taking part in the study, 21 patients were collected for the planned investigations. The inclusion criteria were breast cancer and scheduled axillary dissection after informed consent.
Standardized MR without contrast agent (plain MRI) was followed by intravenous administration of2.6 mg USPIO/kg bodyweight (30-min infusion with filtration). Postcontrast MRI (using the same pulse sequences as for the precontrast studies) was performed 24±36 h after USPIO administration.
The en bloc axillary specimens were precisely marked for comparative studies with the imaging analyses. Deviations from standard axillary dissec- tion were only allowed when suggestive lymph nodes were located outside the anticipated surgical field of axillary exploration.
There was a high drop-out rate among patients in the study, for various reasons.
In the preliminary study only 9 cases (52 lymph nodes) could be evaluable.
Table 1 summarizes the results oflymph node evaluation by two radiologists and the investigat- ing site pathologists.
Preoperative Axillary Lymph Node Diagnosis 19
Using Ultrasmall Particles of Iron Oxide Combined with MRT
S.H. Heywang-Kæbrunner, C. Stets, J. Buchmann, C. Lautenschlåger, F. Gilbert, F. Wallis, 0A. Hupperts, M. Reiser, A. Schauer
Table 1.Histology and size ofthe 52 lymph nodes used in the statistical analysis Maximum transverse diameter of the lymph nodes
Size category <5 mm 5 mm to
<10 mm 10 mm to
<15 mm >15 mm All
Histology Benign 14 10 1 25
Malignant 10 8 5 4 27
All 24 18 5 5 52
The evaluation procedure that gave the best re- sults for combined pre- and postcontrast MRI was based on the following parameters:
· Transverse lymph node diameter £6 mm rather than >6 mm
· Visual assessment of signal change on T1 SE, T2 FSE and PSIF
· Quantitative ratios of signal change SCR to T1 SE, T2 FSE, and PSIF.
On the basis of these parameters, 88% of the lymph nodes were correctly classified with a sensitivity of 85% and a specificity of 92%.
Finally, an accuracy of 87% was obtained when transverse diameter, SCR T2 FSE and vi- sual assessment of signal change on T2 FSE were used as classification parameters.
Among these single parameters, only the transverse diameter and the visual assessment of signal change on T2FSE proved to be signifi- cant discriminators of benign versus malignant lymph nodes (P=0.013 and P=0.007, respec- tively). The most suggestive area of the lymph nodes was determined visually.
The procedures for determination of these areas were as follows:
Comparison of pre- and postcontrast scans revealed differences.
When no suspicious areas could be detected, areas showing either no change or signal inten- sification were chosen.
If neither of these applied, the areas of the lymph node with the widest transverse diameter were chosen.
For evaluation of significance of pre- and postcontrast results areas with iso- or hyperin- tense signal (referred to muscle) on T2 FSE were searched for. If this was not successful, areas with the same characteristics were searched for on FISP.
The signal intensity on the postcontrast T2 FSE scans and the change between pre- and postcontrast T1 SE, T2 FSE and the FISP was recorded as a signal increase, no change, or a signal decrease.
Quantitative evaluations were performed in regions of interest (ROIs):
· Measurement of the representative ROI, care being taken that ROIs were included only on tis- sue type
· As a control, a second ROI sited in the fatty tis- sue around the lymph node was examined, in
order to normalize inhomogeneous reception of the surface coil and differences in the adjust- ment of the MR unit between the examinations.
The following ratios were calculated:
· Signal intensity ratio (SIR) in the precontrast study:
± SIR pre+=(SIR lymph node) pre
· SIR on the postcontrast study:
± SIR post++=(SI lymph node) post
± (SI adjacent fat) post
The histopathological examination was based on systematic work-upof the entire dissection bloc in one site. In the other site the nodal sampling was performed by the surgeon, giving the exact posi- tion of the node to the pathologist.
The following features of each lymph node were included in the statistical evaluation in an SPSS for Windows, vers. 9.0 database:
Precontrast: homogeneity, SIR T1 SE, SIR T2 FSE, SIR FISP visual assessment of signal intensity
Postcontrast: homogeneity SIR T1 SE, SIR T2 FSE, SIR FISP, visual assessment of signal on T2 FSE
The histopathological classification of each single lymph node as
· Completely involved
· Partly involved
· Not involved
The data were entered in a logistic regression anal- ysis. The best results were obtained with the fol- lowing parameters:
· Transverse lymph node diameter (£ 6 mm rather than >6 mm)
· Visual assessment of signal change on T1 SE, T2 FSE and PSIF
· Quantitative ratios concerning signal change SCR on T1 SE, T2 FSE, and PSIF
On the basis of these parameters 88% of the lymph nodes were correctly classified (sensitivity 85%, specificity 92%).
By backward selection the following parameters were eliminated:
· SCR of PSIF
· SCR of T1SE and visual assessment of signal change on T1-SE and on PSIF
Chapter 19 Preoperative Axillary Lymph Node Diagnosis Using Ultrasmall Particles of Iron Oxide Combined with MRT 156
When these conditions were applied an accuracy of 87% was obtained when transverse diameter SCR T2 FSE and visual assessment of signal change on T2 FSE were used as classification pa- rameters.
Among these single parameters only the trans- verse diameter and the visual assessment of signal change on T2 FSE proved to be significant dis- criminators of benign versus malignant lymph nodes (P=0.013 and P=0.007, respectively). Quan- titative assessment of signal change on FSE proved not to be a statistically significant discriminator (P=0.14). Therefore, it was possible to eliminate this third parameter with no deterioration in the results.
For the postcontrast study the following param- eters were available:
· Transverse diameter
· Homogeneity on the postcontrast images
· SIR T1 SE
· SIR T2 FSE
· SIR FISP
· Visual assessment of the signal intensity on T2 FSE
In these evaluations the only parameter that proved statistically significant was the transverse diameter (P=0.002). When this parameter was used an accuracy of 75% was obtained (sensitivity 56%, specificity 96%).
Among these parameters, SIR PSIF performed best, but its discrimination capability did not reach statistical significance (P=0.2).
In the precontrast study, too, only the transverse lymph node diameter proved to have a statistically significant discrimination capability (P=0.003).
When these parameters were used a maximum ac- curacy of 75% was reached (sensitivity 63%, speci- ficity 88%).
Evaluation of Value
The purpose of this first study evaluating the use of USPIO in MRI was to obtain quantitative and qualitative, statistically underpinned parameters which could support standardization of interpreta- tion. In order to allow evaluation on the basis of the lymph nodes only those nodes were to be eval- uated for which histopathological investigations were also available.
There were two points hampering results in the comparative studies that can be incorporated into routine usage:
· Very small lymph nodes, which are sometimes the only ones showing cancer infiltration, can- not be detected.
· Some of the lymph nodes are packed very close together or overlapping.
However, with the current technical conditions, the investigators regard the obtained results as ªfairly representativeº of most of the axillary lymph nodes in the eligible patients.
The study is not ªpatient based.º However, at least a sufficient number of lymph nodes of all sizes was included in the evaluation. Therefore, the results must be accepted as a ªtest setº in order to determine the value of the various potential diag- nostic parameters and to find thresholds for im- portant parameters and the optimum effective combination of these parameters.
The following points are of interest for further approaches:
· Only a few parameters are necessary for optimal accuracy.
· Quantitative evaluation of the signal change (even on FSE) is inferior to visceral assessment of the signal change and therefore not even nec- essary.
· Lymph node homogeneity seems not to be an important discriminator in any type of study (precontrast, postcontrast or combined).
· A clear advantage over plain MR in terms of ac- curacy, especially concerning sensitivity, is ob- vious (12%).
· Comparison of post- and precontrast study re- sults showed increased specificity at cost of sen- sitivity. Thus, overall accuracy was not im- proved.
Overall, with all facts taken together, in cases in which imaging detects macroscopic lymph node involvement with sufficiently high specificity im- mediate axillary dissection could be chosen as the more appropriate, more cost-effective and less time-consuming method.
Since the combined analyses provide good spec- ificity and allow further development, the combi- nation may become a valuable adjunct to support diagnostic and surgical strategies at an early stage.
In a further preliminary study of the two sites carried out by the same research groups the same
pulse sequences and field strength as in the precur- sor study (pre- and postcontrast MR) were used.
T1-weighted and T2-weighted SE and 3D PSIF obtained in 88 lymph nodes (39 benign, 49 malig- nant) in 14 patients were histopathologically corre- lated and evaluated visually and quantitatively in a double-blind controlled investigation. Combina- tions ofdiagnostic parameters were compared and evaluated in logistic regression analysis.
The best accuracy (85%: sensitivity 77%, speci- ficity 92%) was obtained with transverse lymph node diameter (>6 mm): visual assessment and signal intensity ratios (SI lymph node/SI fat) of T2-weighted SE and 3D-PSIF before and after CM and the (SI ratio after CM/SI ratio before CM).
Without quantitative parameters accuracy was 81%. Precontrast data alone yielded 59% accuracy, postcontrast data alone 62%.
The authors summarize from these data that both pre- and postcontrast scans are needed. In addition, CM improves accuracy significantly and quantitative evaluation improves accuracy slightly over visual evaluation.
In a phase III study carried out by Taupitz et al.
(1999) ofthe aforementioned international research group, Sinerem given as an i.v.-injected contrast me- dium for MRI was tested for lymph node imaging in patients with suspected breast cancer.
In 35 patients breast surgery and axillary lymph- adenectomy were performed within 10 days of each other.
Plain MRI examination ofthe axilla was per- formed before an i.v. Sinerem infusion.
The contrast medium was infused at a dosage of 2.6 mg Fe/kg within 30 min. Patients had a second comparative MRI within 24±36 h after the infusion (Taupitz et al. 1999). MRIwas performed at 1.0 or 1.5 T.
All patients were examined with three sequences before and after Sinerem application:
Axial T1-weighted Se, axial T2-weighted SE or T2-weighted fast SE, axial or coronal conventional T2-weighted GRE or T2-weighted 3D-PSIF. Lymph node assessment was based on
· Lymph node size for plain MRI
· Signal variations for Sinerem-enhanced MRI.
The results were correlated with histopathological findings.
The preliminary results recorded by the group are summarized in Table 2. For all patients, with plain MRI there were 7/9 true-positive and 7/5 false-positive results, while with Sinerem-enhanced
MRI there were 12/15 true-negative and 1/4 false- negative results.
When a lymph node was considered nonmeta- static (short axis <10 mm) on plain MRI or show- ing a signal decrease in at least one sequence on Sinerem¾ enhanced MRI, plain MRI and Sinerem enhanced MRI showed
sensitivity of50% and 64% and specificity of 75% and 94%,
for the detection of lymph node involvement, re- spectively.
The conclusion drawn from this phase III study agrees with the judgment already presented in the main chapter insofar as Sinerem-enhanced MRI performs better than MRI alone, according to size criteria alone, in identification of metastatic lymph node involvement.
This seems to be an important development as far as the four imaging-systems are concerned, which can never totally exclude metastatic pro- cesses. This clear statement can be given without any restrictions, because even with all our experi- ence it is difficult to exclude metastatic processes.
Such early metastatic processes can be detected by combined histological and immunohistochem- ical microscopically based examinations in serial sectioning the nodes. But in cases, in which Siner- em oxide labeling indicates already a metastatic node process already preoperatively decision for total axilla revision (level I and II) can be made.
In addition Sinerem labeling in optimal perfor- mance could also help to detect parasternal lymph node involvement, which is in relation to addi- tional parameters important for decision making ofadjuvant therapeutic regimens.
Chapter 19 Preoperative Axillary Lymph Node Diagnosis Using Ultrasmall Particles of Iron Oxide Combined with MRT 158
Table 2. Results when Sinerem by i.v. injection as contrast medium for MRI was tested for lymph node imaging in pa- tients with suspected breast cancer (Taupitz et al. 1999)
Lymph node detection ac- cording to MRI and after Sinerem infusion
Plain MRI After Sinerem in- fusion Lymph node detection
according to MRI and after Sinerem infusion
N=234 lymph nodes
N=244 lymph nodes Total removed by surgery N=501a
lymph nodes Possible comparison of
lymph nodes MRI/
histology
N=144 lymph nodes
aLymph nodes not detected were mostly smaller than 9 mm
Lymph Node Staging by Means of USPIO in Lung Cancer Patients
A. Schauer
In a first multicenter clinical trial of USPIO in the evaluation of mediastinal lymph nodes in patients with primary non-small-cell lung cancer, Nguyen et al. (1999) tested 18 patients with primary lung malignancies and suspected regional lymph node metastases.
After the initial performance of MR sequences, each patient was evaluated for the number and lo- cation of the mediastinal lymph nodes, their homogeneity based on the nodal signal, and possi- ble change of the MR signal postcontrast applica- tion.
The lung resections and sampling of the MR- identified nodes or node groups were performed 1±35 days after contrast MR imaging. There were 27 lymph nodes or node groups available for his- topathological correlation.
The combined MR imaging had a sensitivity of 92% and a specificity of 80% in confirming me- diastinal lymph node(s) involvement. The authors conclude that combined MR imaging can support mediastinal lymph node staging.
Such preoperative evaluations and the knowl- edge gathered from them can support decision making about surgical treatment even in the treat- ment planning phase and possibly also in the plan- ning of additional adjuvant therapeutic efforts.
References
Mohr U, Weiûleder R (1996) Lymph node diagnosis with imaging methods. An overview with special reference to recent developments in the area of contrast media. Z Lymphol 20:15±20
Nguyen BC, Stanford W, Thompson BH, Rossi NP, Kernstine KH, Kern JA, Robinson RA, Amarosa JK, Mammone JF, Outwater EK (1999) Multicenter clinical trial of ultra small superparamagnetic iron oxide in the evaluation of mediastinal lymph nodes in patients with primary lung cancer. J Magn Reson Imaging 10:468±473
Stets C, Wallis F, Pickuth D, Lautenschlåger C, Murray AD, Heywang-Kæbrunner SH (2000) Iron oxide enhanced in- travenous MR-lymphography in breast cancer: compara- tive value of precontrast, postcontrast and combined evaluation (abstract). RSNA, December 2000, abs 878 Stets C, Brandt S, Wallis F, Buchmann J, Gilbert FJ, Hey-
wang-Kæbrunner SH (2002) Axillary lymph node metas- tasis: a statistical analysis of various parameters in MR with USPIO. J Magn Reson Imaging 16(1):60±68 Taupitz M, Wallis F, Heywang-Kæbrunner SH, Thibault F,
Gilles R, Tardivon AA (1999) Axillary lymph node MR imaging with Sinerem in patients with suspected breast cancer (abstract). RSNA, December 1999, abs 1172
