| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Department of Radiology, Imaging Diagnostic Systems, 6531 NW 18th Court, Plantation, FL 33313
e-mail: encm{at}hotmail.com
I would like to congratulate the authors of the article "Benign versus Malignant Breast Masses: Optical Differentiation with US-guided Optical Imaging Reconstruction" in the October 2005 issue of Radiology (1). I believe this article heralds an important advance in our ability to distinguish benign from malignant lesions in the breast without recourse to biopsy.
The aim of using an optical technique for this purpose is essentially to provide the clinician with nonmorphologic data that will allow him or her to improve the triage function of mammography and reduce the presently very high negative biopsy rate. My first concern, therefore, with this study is that additional biopsies were actually performed, including one of the contralateral breast to "provide a better reference," as the authors state, "for tissue homogeneity." Obviously, if the aim is to reduce the number of biopsies, reference biopsies cannot be used in the clinical environment. Could the authors explain how they propose to handle this, or were the reference biopsies necessary only for this particular study?
My second concern is the authors' statement that "invasive cancers revealed a more than twofold greater total hemoglobin concentration" (than benign lesions). While this is true if we look at the mean values only, mean values are of limited value in individual clinical practice and, if we consider the values reported in this article, the hemoglobin concentration for five of the eight cancer cases ranges from 98 to 115 µmol/L (mean, 106 µmol/L) and the concentration for 12 of the benign lesions ranges from 80 to 130 µmol/L (mean, 105 µmol/L). In clinical practice, therefore, this two-times difference in hemoglobin concentration would frequently not occur.
The authors state that "no false-negative results were found"; however, it would appear that the cutoff point of 95 µmol/L for differentiating between benign and malignant was selected retrospectively so that no false-negative results would be obtained. It is by no means certain that this same 95 µmol/L figure would not produce false-negative findings in actual practice. Also, it would appear that the separation into true-positive and true-negative results is highly dependent on the accuracy and repeatability with which the hemoglobin concentration can be determined. A difference of 10 µmol/L in this study (from 95 to 105 µmol/L) would mean missing three of eight cancers. How did the authors determine (or can they reference an article that determines) the accuracy and reproducibility of optical measurements of hemoglobin in the human breast?
My third comment concerns the authors' statement that "no significant differences were demonstrated among findings in the benign group." This does not accord with the known wide variation in vascularity of benign lesions or with the authors' own data. Their figure 3 shows maximum hemoglobin concentrations ranging from 20 to 120 µmol/L for fibroadenomas and from 15 to 95 µmol/L for complex cysts. These data accord very well with our own findings by using an optical imaging technique known as computed tomographic (CT) laser mammography (2). The authors are apparently not aware of this technology, since they state that "because of intense light scattering, optical tomography alone has not been widely used in clinical studies. The data in the published literature have been limited to feasibility studies or case reports."
It is important to dispel this widely held belief that optical tomographic imaging is not feasible in the human patient because scattering will lead to poor spatial resolution. This belief is erroneous, because it does not take into account the fact that optical imaging provides excellent contrast resolution of angiogenesis and (as the authors indicate) that the volume of angiogenesis associated with any tumor is usually much larger than the tumor itself (eg, a lesion that is 3–4 mm in size on a mammogram is usually associated with an angiogenesis volume of 2–6 cm3), so that it is not necessary to have high spatial resolution to visualize angiogenesis in the human patient (2–4). The optical system described in these reports (CT laser mammography) has been employed in over 5000 cases to date in Europe, the Middle East, and Asia.
Despite the practical questions I have raised, I believe the authors are very much on the correct track. Extensive testing of the most recent advance in breast imaging (the American College of Radiology Imaging Network study of digital mammography) has served to confirm that high-spatial-resolution depiction of pure morphologic data is not in itself sufficient to provide either high sensitivity (44% of all cancers in this nationwide study were missed) or high specificity (75% of all biopsy results were negative), confirming that material improvement in mammography cannot be achieved simply by better visualization of pure morphologic data but requires the addition of functional information of the type obtained by means of optical technology (5).
 |
References |
|---|
 TOP References References |
|---|
and
Scott Kurtzman, MD
* Bioengineering Program, Electrical and Computer Engineering Department, University of Connecticut, 371 Fairfield Rd, U2157, Storrs, CT 06269-2157
e-mail: zhu{at}engr.uconn.edu
Department of Radiology, Hartford Hospital, Hartford, Conn
Department of Surgery, University of Connecticut Health Center, Farmington, Conn
We thank Dr Milne for his interest in our article and appreciate his positive comments. Dr Milne raises three concerns about our study, and we will address each of them below.
The first concern is regarding the use of the contralateral breast as a "reference." In our study, we acquired ultrasonographic (US) images and optical data at the lesion site, a normal symmetric region of the affected breast, and a normal region of the contralateral breast in the same quadrant. The normal sites are used to estimate background optical properties used for imaging reconstruction. We did not perform additional biopsies at any normal sites and the only biopsy, or biopsies if multiple lesions were involved, was the lesion (or lesions) requested by patients' physicians. Unfortunately, we did not notice the change in meaning in a sentence when we reviewed the copyedited version of our manuscript. An erratum was published in the May 2006 issue of Radiology to clarify the sentence.
We share the other two concerns with Dr Milne, and this is exactly why we provided both group average (figure 2) and individual data (figure 3). When we state that "no significant differences were demonstrated among findings in the benign group, but a more than twofold higher total hemoglobin concentration (Fig 2) was found in the malignant group," we were discussing or referring to figure 2 of group averages. We agree that individual values are very important for clinical practice; however, group averages provide the statistics in general. We did not find any statistical difference in hemoglobin concentration among averages of all of the benign group; however, we found a more than twofold higher average in the malignant group than among the averages of the benign group.
To the best of our knowledge, our pilot study is the first one using a priori US lesion location to guide reconstruction of quantitative total hemoglobin distribution for distinguishing malignant from benign lesions among a moderate patient population. Retrospective evaluation by using a threshold is necessary to quantitatively classify populations with malignant and benign results in this pilot study. Our data can be considered as a training set for future large-scale prospective studies. Recently, there have been several pilot studies using projection images and human observers to establish diagnostic criteria (1,2). In another recently published pilot study (3), the light reflection measurements have been used to quantify the incremental blood volume and blood oxygen saturation of benign and malignant lesions with the use of retrospectively selected thresholds.
In the past, we have reported several phantom studies to quantify the accuracy of image reconstruction of optical absorption coefficients (4,5). Total hemoglobin concentration is a computed value from absorption coefficients of two optical wavelengths. In one study (5), small (about 1 cm) absorbers of both low (0.08–0.1 cm–1) and high (0.18 cm–1) absorption contrasts were used to test the reported US-guided optical imaging algorithm. The reconstructed values of small absorbers of both low and high contrasts can reach 76%–102% of the true values, depending on target depth. In vivo evaluation of hemoglobin is not an easy task. However, since hemoglobin level is closely related to tumor microvessel density, we have correlated total hemoglobin concentration with microvessel density count and obtained a reasonable correlation coefficient of 0.64 (P < .05) from another small-scale clinical study (6).
We wish to emphasize our important finding of hemoglobin distribution versus cancer size. In our reported study and in another published clinical study (7), the invasive carcinomas were approximately 1 cm in size, with one of 2.2 cm in size. The hemoglobin levels were quite high and masses were isolated compared with the diffused patterns of most benign lesions. In another study (6), of six large cancers ranging from 2 to 4 cm in size, the observed hemoglobin distributions are quite heterogeneous due to complex angiogenesis distributions in larger cancers. Since large US-visible cancers are readily diagnosed with US, optical tomography offers the potential to map tumor vascularity and tumor hypoxia. These indexes can be obtained before and during chemotherapy to assess treatment efficacy.
As for the study cited by Dr Milne, there were no published clinical trial results or partial results by the time our manuscript was submitted (July 2004) and in revision (September 2004). The article cited by Dr Milne (2) was published in June 2005, while our manuscript was accepted in December 2004.
The last comment is related to optical scattering versus spatial resolution. It is true that intense tissue scattering leads to poor spatial resolution of optical tomography. In addition, the intense scattering also causes lesion location uncertainty in reconstructed images (4,8,9). Another important issue is that the quantitatively reconstructed light absorption coefficients are much lower than the expected values (4,5). This is the main reason that we use US-localized lesion to guide the optical imaging reconstruction and improve the light quantification accuracy.
|
References |
|---|
|
TOPReferencesReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
