Breast Thermography: History, Theory, and Use

 

The National Cancer Institute estimates there will be 226,870 women diagnosed with breast cancer in the United States in 2012.¹ There are now estimated to be 2.6 million breast cancer survivors.² Early detection and treatment of breast is the best current strategy for reducing the morbidity and mortality of the disease. An ideal screening method would be one that is sensitive enough to detect breast cancer early, specific enough to differentiate malignant from benign lesions, easily accessible to the general public, financially feasible, and unlikely to cause harm to the patient. Currently, mammograms are the only U.S. Food and Drug Administration (FDA)–approved standalone diagnostic tool for screening use in the general population. However, like any imaging procedure, mammograms have limitations, including fairly low sensitivity, particularly in those with dense breasts.³ There is also a fairly high false positive rate, with up to 50% of women getting annual mammograms for 10 years, which leads to unnecessary biopsies of benign lesions in 25% of these women.⁴ Lastly, there is the small but measurable risk from the gamma radiation when used repeatedly, particularly in those of a younger age.⁵ Even with these limitations, mammograms are the best imaging tool for detecting breast cancer early, and certainly many women who have had breast tumors found and removed due to mammography screening are advocates of the benefits of such early detection. In terms of precision, cost, access and risk, mammograms are better defined than any other screening method.

 

It is essential to distinguish a standalone screening tool, such as mammography, with those that are adjunctive such as ultrasound, magnetic resonance imaging (MRI), scintimammography, thermography, and electrical impedence, all of which are FDA-approved. In a 2004 systematic review of screening techniques, only ultrasound, MRI, and mammography had sufficient data to determine their utility as screening tools.⁶ Not only thermography, but CT scanning, magnetic resonance spectroscopy (MRS), scintimammography, electrical impedance, infrared spectroscopy, light scanning, and positron emission tomography were all excluded due to lack of rigorous data. Today, ultrasound and MRI are the most common adjuncts to mammography in breast cancer imaging, and both have been shown to have predictable sensitivity and specificity based on large, randomized trials.

 

Interestingly, long before the approved use of ultrasound or MRIs, breast thermography was a promising imaging technique. Breast thermography, as the name implies, renders an image of the breast based on temperature differences. Rather than a morhpological depiction of the breast tissue, thermography renders a functional image as it is seen in a change of temperature at the skin surface. The FDA approved thermography as an adjuctive tool in the assessment of breast masses in 1982. Despite its early promise, breast thermography has fallen far behind in the race for validated breast imaging procedures. As a screening tool, there is very little to substantiate the claim that it is capable of detecting cancer at the most opportune time for treatment—in its early stages. Studies suggest it may be useful as an adjunct to mammograms, but lack of standardization and large trials preclude its widespread use. 

 

The theory of breast thermography begins with the premise that breast tissue free of any abnormal processes has a predictable emanation of heat patterns on the surface of the skin. When physiological processes, such as vascular disturbances or inflammation, are present, there is a disruption of the normal pattern, which can be captured through sensitive equipment.⁷ In the beginning, measurements of such temperature variations were primitive, even including a direct probe on the skin. Today, breast thermography equipment can detect small variations in infared emanations, corresponding to skin temperature differences of as little as 0.025ºC.⁸

 

Thermography is a noninvasive technique that does not employ any radiation. In a temperature-controlled setting, patients disrobe to the waist and allow the surface of the breasts to cool to room temperature (18–22ºC) for 10–15 minutes. With arms raised, the woman has 3 images taken (1 anterior and 2 lateral views). An optional “cold challenge” may be done, in which the patient places both of her hands into a 10ºC water bath for 10 minutes. The premise of the cold challenge is that the abnormal physiology of the highly permeable vascular network around tumors does not respond normally (ie, vasoconstrict) with the cold stimulus, while normal breast tissue does. In theory, this would enhance the abnormally vascularized areas, although this entire premise has been questioned in one review of the literature done in 2004.⁹

 

Dr. Ray Lawson published “Implications of Surface Temperatures in the Diagnosis of Breast Cancer” in 1956. His paper included a picture of a large breast mass demonstrating temperature variation with surrounding breast tissue.¹⁰ In the same paper he reported on a series of 26 women with proven breast cancers, for whom "the average detectable temperature rise in either the area of the tumour or the ipsilateral areola was 2.270ºF.” While the techniques he used are now outdated, the hypothesis set off more than 2 decades of research into methods to improve breast thermography as a useful screening tool. It is important to note that during this time, there was no other screening procedure widely available, and the use of xerography (mammography) was being researched simultaneously. 

 

The fledgling field of thermography appeared promising, as the benefits of early detection of breast cancer were already quite clear. Many letters and opinions were exchanged touting the virtues of the technology’s potential,^11,12 and research and development continued alongside that of other imaging techniques.^13,14,15,16 However, mammography was clearly emerging with superior sensitivity and specificity over thermography,¹⁷ so one could say it got an early lead in the race for the best imaging technique. There were also better designed and larger randomized controlled trials on mammography, something that thermography is lacking even today. 

 

Whereas the mammogram renders an anatomical representation of the breast, a thermogram provides an image based on physiology, and interpretation of those images was highly subjective, thus inconsistent. To be clear, even in the earliest days, thermography held the promise of indicating a general abnormality, not necessarily differentiating a benign from malignant condition. In these early years of development, attempts to use thermography as a prescreening tool to indicate which women should go on to get a mammogram was intended to limit the exposure of x-ray radiation in the screening process. While the theory was logical and its intent admirable, thermography never achieved the sensitivity and specificity necessary to be used as a prescreening tool.

 

One large study in 1972 was particularly instructive in the use and limitations of thermography. Dr. Harold Isard published a paper on the accuracy of screening in 10,055 women using clinical examination, mammograms, and/or thermograms separately and together over a 4-year period (1967–1970).¹⁸ Fifty-six percent (n=5,662) of participants had symptoms such as a palpable mass, nipple discharge, or pain. Forty-four percent (n=4,393) of participants were asymptomatic, with many having a family history of breast cancer. Cumulatively, there were 306 histologically confirmed breast cancers (270 in the symptomatic group and 36 in the asymptomatic). In symptomatic women, clinical breast exam, mammogram, and thermogram correctly diagnosed 82%, 85%, and 72% of the lesions respectively. When thermography was added to mammography, the accuracy increased from 85% to 92%. In the asymptomatic group, there were 36 cancers histologically confirmed. Thermography correctly diagnosed 61% of these women, while mammograms accurately predicted 83%; combining both imaging techniques rendered an 89% accuracy rate. In this study the use of mammography was clearly superior as a standalone technique, and thermography appeared to have some additive benefit. The paper is a seminal work in the body of thermographic publications, as Isard outlined distinct criteria for the interpretation of normal and abnormal thermographic images, an important contribution in creating more standardized and replicable interpretation of the results. The state of thermography was perhaps best captured by Isard himself in his introduction to the paper: “The thermogram at this stage of development of the discipline of thermography can no more by itself differentiate a benign from malignant condition than can the temperature recording by the oral thermometer differentiate pneumonitis from a necrotizing neoplasm.”

 

In that same year, a publication by Nathan and colleagues of 359 women, most of whom had breast symptoms, looked at breast thermography as a screening procedure.¹⁹ Overall, there were 195 abnormal thermograms and 164 normal thermograms. Of these 195 abnormal results, only 27 were found to have cancer, 53 had benign lesions, and 115 had no organic disease at all. Of the 164 patients found to have a normal thermogram, 7 had cancer, 41 had benign lesions, and 116 had no organic disease. This rendered a false negative rate of 29%, leading the authors to conclude, “mammary thermography is of no practical value in the differential diagnosis of symptomatic mammary disease.” 

 

In an attempt to more clearly define the benefits of early detection of breast cancer, the Breast Cancer Detection and Demonstration Project (BCDDP) began in 1973 under the auspices of the American Cancer Society (ACS) and the National Cancer Institute (NCI). There were 27 BCDDP centers around the country. Each center offered voluntary screening to women between the ages of 35 and 74. Screening consisted of medical history, physical examination, mammography, thermography, and instruction on self breast exams. In a publication of the NCI in 1978, a panel relayed the recommendation of a consensus group that had reviewed the available data on thermography to date and found “that although there is no known harmful effect from thermography, there are no scientific data supporting its value as a routine breast cancer screening technique under present conditions of general use. They [the consensus group] strongly suggested that research be carried out to improve thermographic techniques and to determine its role in screening. It was recommended that thermography be discontinued as part of the routine BCDDP screening process except in those centers where proficiency is available to justify further clinical investigation under appropriate research design.” All BCDDP centers continued to offer thermography, but clinical directors were requested to design appropriate research methods to assess and determine its utility.²⁰

 

The above recommendation from the NCI was based largely on the publication by Stephen Feig, MD, and colleagues then at Thomas Jefferson University Hospital in Philadelphia. In 1977, Feig reported on the use of clinical exam, mammography, and thermography in 16,000 self-selected women (40–64 years old) who participated in the BCDDP project at the hospital.²¹ He found that of the 139 biopsy-proven malignancies in the 16,000 women screened, mammography detected 78% (109), clinical exam 55% (76), and thermography 39% (54). In his publication mammography was clearly superior to thermography and clinical exam in detecting occult disease, while thermography was more likely to correlate with clinically palpable tumors. This large study turned the tide of emphasis in breast cancer screening toward further refinement of mammograms in lieu of thermography. While the BCDDP pathology techniques were criticized even within the NCI,²⁰ there was no disputing that this large study was in agreement with the vast majority of the smaller datasets prior to 1977.  With this influential publication by Feig and the NCI recommendation, thermography lost momentum in the competition for both attention and research dollars from the medical community. 

 

While focus in radiology turned mostly to mammography, small pockets of investigators continued to assess and refine the use of themography in breast imaging. For example, in 1988 Isard and colleagues published a novel use of thermography as a prognostic tool.²²  He devised aprognostic classification for thermographic staging of breast cancer and applied this to a cohort of 20 women treated before 1980 at Albert Einstein Medical Center. His conclusion: “The thermographic scoring system clearly shows shorter survival for patients with poor thermographic prognostic factors, 30% surviving at 5 years and only 20% at 10 years compared with overall survival of 80% at 5 years and 70% at 10 years.” This was a very small population of 20 patients and to date this work has not been validated or duplicated by any other studies, rendering his results interesting but not clinically useful.

 

A 1980 study looked back at 1,275 women who had had a moderately abnormal thermogram 5 years before. All the women had benign breast disease diagnosed via physical exam, mammogram, ultrasound, and possibly biopsy. This study showed that more than a third of these women went on to develop histologically confirmed cancers in the ensuing 5 years.²³

 

However, in a separate publication by Lloyd Williams and colleagues published in 1990, thermograms failed to have any predictive value as a screening tool in a large population of both physician-selected and volunteer-based women (10,238 total participants, ages 40–65).²⁴ All of the women underwent physical examination and thermogram. If the thermogram was abnormal in premenopausal women, it was repeated in the second week of the menstrual cycle. Of the 2,681 women with an abnormal thermogram, 201 reverted to normal upon repeat imaging.  If clinical examination or thermogram indicated an abnormality, the patients underwent mammography. At the time of screening, 59 women were found to have breast cancer, 19 of whom had symptoms of “skin tethering, a discrete lump or nipple abnormality.” Of those 59, 61% had a positive thermogram. At 5 years follow-up (96.9% of participants were followed), 60 women had developed breast cancer.  Thermograms were positive in 28.3% of these women and negative in 71.6%, which was comparable to the 26% positive an 74% negative thermograms in the women who did not go on to develop cancer. This study essentially reaffirmed that while thermograms were capable of detecting larger, discrete masses, it failed to have any value as a screening tool in detecting occult tumors.

 

Since the study conducted in 1977 by Feig and colleagues that led to a sea-change in the direction of breast imaging, a number of technological advances in thermography have occurred leading to better imaging and more objective analyses of the thermographic images.  Despite the advances, the summation of the evidence for thermography is still not strong enough to recommend its widespread use. 

 

One advantage we have today is computer modeling, which allows for rapid development of so many medical techniques. Computer simulation models may offer a means of projecting “what if” scenarios that may advance our understanding of how to interpret various thermographic patterns.²⁵ There are many models currently being investigated to improve the precision and sensitivity of infared imaging.  One has looked into the use of neural network patterns along with temperature changes to better predict the presence of a tumor.²⁶ Models using more current breast thermography techniques show that, theoretically, masses less than 0.5 cm may be detected, but only if close to the surface of the skin.²⁷

 

A general advancement to modern thermographic techniques is the use of digital infared thermal imaging (DITI), which allows the image to be digitally captured and manipulated. An early trial of DITI in 1998 by Keyserlingk and colleagues retrospectively reviewed the ability of clinical exam, mammography, and thermography to detect a series of 100 cases of ductal carcinoma in situ, stage I or stage II disease.²⁸  High-resolution digital thermography was used and found to have a 17% false negative rate. The false positive rate was 19% in a concomitant series of 100 benign breast biopsies. In keeping with the previous research, thermography was found unsuitable as a standalone screening technique; however, thermography did improve the sensitivity of mammographic findings, in this study from 85% to 95%. 

 

More recently, in 2011, a small study of 63 women with suspicious breast lesions were evaluated by DITI and subsequently underwent surgical excision or core biopsy of the lesions. The sensitivity of DITI in detecting biopsy-proven malignancy was 25%, and the specificity was 85%.²⁹ An earlier study in 2008 by Arora and colleagues performed DITI in 92 patients for whom a breast biopsy was recommended based on prior mammogram or ultrasound.³⁰ They reported that DITI identified 58 of the 60 malignancies subsequently confirmed as cancer histologically, resulting in 97% sensitivity and 44% specificity. These large fluctuations in the sensitivity and specificity, despite using modern DITI techniques, underscores the need for larger trials with better standardized protocols to definitively determine the sensitivity and specificity of thermography in breast imaging.

 

 

 

The best means of reliably interpreting thermographic patterns continues to undergo refinement. In an attempt to improve on the analysis of results obtained from modern DITI, Wishart et al compared 4 different methods for digital breast imaging analysis of 106 biopsy specimens from 100 women who were going to obtain biopsies of mammographically identified lesions.³¹ Before biopsy, thermograms were obtained and analysis of the images was done in a blinded fashion using 1 of 4 methods (the details of which are not essential to understand):  sentinel screening report, sentinel artificial intelligence (neural network), expert manual review, and NoTouch BreastScan software (an artificial intelligence program). Overall, sensitivity was best for expert manual review (78%), and NoTouch software (70%). NoTouch software analysis performed better in women under 50 years of age, with a sensitivity and specificity of 78% and 75% respectively. This study also confirmed that combining thermography with mammography improved the sensitivity of screening, in this case to 89%. Such studies show promise, but again, must be replicated and validated on a larger scale to begin to change clinical practices.

 

Later generations of thermography techniques are clearly improving upon the older, less accurate techiniques.³² Technological advances continue to improve the accuracy of breast thermography, particularly for determining small tumors and those deeper in the breast.³³  However, the methodology still needs to be standardized, validated, and compared head to head in clinical studies using current ultrasound and MRI techniques as comparative adjuncts to mammography.

 

Currently the Marseille system of classification is used to categorize the results of a thermogram.⁸

                 

TH-1        No unusual features; normal breast tissue

TH-2        Area(s) of increases in heat that are responsive to the cold challenge

TH-3        Area(s) of atypical increases in heat that are not responsive to the cold challenge

TH-4        Area(s) of abnormal increases in heat that are not responsive to the cold challenge

TH-5        Area(s) of severely abnormal increases in heat that are not responsive to cold challenge

 

According to the National Cancer Institute, mammograms have an absolute mortality benefit of approximately 4 out of 10,000 women screened annually between the ages of 40–49, and 5 out of 1,000 women screened annually over the age of 50.³⁴ The greater advantage with advancing age is due to the increased risk with age. A review of the research on the utility of screening mammograms led by the US Preventative Services Task Force (UPSTF), an independent panel of experts, concluded screening mammograms are best evidenced for use in women between the ages of 50–75, on a biennial basis. This recommendation in 2009 led to public outrage, political posturing from both parties, and various organizations decrying the adoption of the recommendations.³⁵ The recommendation to begin mammographic screening at 40 years of age and continue it annually was reiterated by the American Medical Association, the American Cancer Society, the American Congress of Obstetricians and Gynecologists, the Society for Breast Imaging, and the American College of Radiology (ACR). A public awareness campaign is still in effect using the slogan trademarked by the ACR, “Mammography saves lives,”³⁶ followed by “One of them may be yours.” The highly emotional rhetoric in the mammography debate is not surprising, given the vastly different outcomes of early versus late diagnosis. 

 

The continuing debate over the timing and frequency of mammography as a screening tool is instructive to the discussion of breast thermography. The amount of data we currently have on the use of thermography, either as an adjunct or standalone screening method, pales in comparison to the plethora of data available on mammography. This implies that the data on thermography is, in short, not adequate to provide definitive guidance. However, some integrative medical practitioners currently employ the use of breast thermal imaging to manage a number of clinical queries: 1) to evaluate a patient’s risk of developing breast cancer, 2) to monitor perimenopausal women initiating hormone replacement therapy, 3) to evaluate a suspicious area of the breast, and/or 4) as a screening tool for women who either cannot or will not get a mammogram. However, such uses demonstrate considerable misgivings about the use of breast thermography, and a review of the literature shows that such uses are not evidence-based. 

 

Proponents of breast thermography postulate that the thermal image of the breast may be detecting carcinogenic changes (such as locally increased disordered vascularity) 5–10 years prior to the development of a detectable lesion. In 2000, Jonathan F. Head of Medical Thermal Diagnosics in Baton Rouge did a retrospective look at 3 different groups of patients who underwent breast thermography beginning in 1973 at the Elliott Mastology Center.³² All participants had undergone thermography 1 year or more before their breast cancer diagnosis. Group I was composed of 126 patients who had died from breast cancer, group II was composed of 100 patients who had been diagnosed with breast cancer, and group III had been diagnosed with a variety of benign mastopahies. In a retrospective look, abnormal thermograms were found in 88% of group I, 65% of group II, and 28% of group III. While this trend is impressive, the report must be interpreted with some skepticism given the funding and affiliation with a vested thermographic entity and the lack of peer review of the publication.

 

One must keep in mind that although thermography is advancing in its techniques and continues to be developed as an adjunctive tool, advances are also being made in other breast imaging techniques as well. Improvements in breast imaging continue to move forward, with mammographic techniques, ultrasound, and MRI all undergoing intense innovations to maximize their accuracy.³⁷ In addition, many other imaging methods are currently in development alongside these commonly used techniques, including nuclear medicine imaging (eg, radioimmunoscintigraphy, positron emission mammography), the 360-degree CT scan, diffraction techniques (eg, diffraction-enhanced imaging, small-angle X-ray scattering), diffuse optical imaging, and electrical impedence scanning).³⁷ As of the writing of this paper, the FDA Radiological Devices Panel has recommended the FDA approve the use of a specific automated ultrasound scanning system (U-systems somo-V®) for use in asymptomatic women with dense breasts.

 

In summary, the evidence indicates that breast thermography is not suitable as a standalone screening technique. While current technological advances in imaging and analysis may eventually deliver on the promise of thermography to detect occult or even precancerous lesions, there is no good evidence to date indicating that this use is appropriate. The high false positives from thermography can lead to unnecessary psychological stress and should be included as “harm” when assessing the pros and cons of thermography.  Conversely, patients who undergo thermography without concomitant mammography and go on to receive a false negative thermogram are even further harmed by a possible delay in diagnosis. Ultimately, there are no large randomized, controlled trials using standardized protocols that can guide us to even gauge the actual risk/benefit of thermography in an evidence-based manner. The preliminary data suggest that its use as an adjunct with mammography may increase the sensitivity of mammographic imaging, particularly in younger women. Certainly, as an add-on to mammography, there is less risk to the patient as the standard of care (mammography) is included. Ultimately, the paucity of published clinical research supporting the use of breast thermography for the detection of breast cancer or carcinogenic changes is a major limiting factor for effective clinical application. As advances in breast thermal imaging arise, they should be appropriately evaluated in clinical trials for their most promising clinical applications, as well as a greater understanding of their limitations. Improvements in technology and interpretation will continue to emerge regarding thermography, and it is important to remain open-minded, as the possibility of validation or invalidation of thermography as a screening tool in the future are both still on the table.