Thermography

We were very surprised to see that Sterns et al.1 had results that contradicted those presented by our group which demonstrated an association between three growth rate-related prognostic indicators for breast carcinoma and findings from infrared imaging.2 Careful review of the Sterns et al. article1 showed that they performed the infrared imaging with a different and outdated technology, did not quantitate actual growth rate with serial measurements from mammography, and looked at only one of the three growth rate-related parameters reported by our group2 to be associated with abnormalities found in infrared images of the breast. In reviewing the two articles by Sterns et al.1, 3 on contact thermography, we noted that they reported that 56% of the breast carcinoma patients from the initial 214 patients who were screened with thermography had an abnormal breast thermogram, whereas only 19% had abnormal breast thermography when 420 women were enrolled in the study. This inconsistency is due to the fact that in the initial study the patients with equivocal thermal patterns were considered abnormal, whereas in the recent expanded study equivocal patients were considered normal. This clearly demonstrates that the authors are not sure what level of abnormality is really significant when attempting to relate infrared images to growth rate and growth rate-related parameters. In our study,2 in which telethermography was used, 65% of the breast carcinoma patients had abnormal thermograms (including both the small number of patients with slightly abnormal thermograms and all the patients with definitely abnormal thermograms), and Isard et al.4 found that 54% of their patients had either PF2- or PF3-level abnormalities. In both of these studies, the presence of an abnormal breast thermogram had a significant impact on survival. It is not appropriate to compare contact thermography results, produced by a technology that is over 20 years old and was not used in either our study or that of Isard et al.,2, 4 with the results from infrared telethermography (a newer, more sensitive technology with much better image quality). It appears that the superior image quality of infrared telethermography may be necessary to demonstrate that abnormal infrared images of the breast (asymmetric hot spots, global heat, vascularity, areolar/periareolar heat, or edge sign) are related to survival and growth rate-related prognostic indicators. Our emphasis at present is to improve our sensitivity and image quality further by using second-generation infrared imaging technology, which has previously been used only by the military. The second-generation infrared imaging system that we have been evaluating is more sensitive than scanning telethermography units and produces a better image, with a 256 × 256 element focal plane array sensor. In addition, we are currently developing methods of analyzing the infrared images of the breast with computerized image analysis software that compares the heat pattern of one breast with that of the other to delineate the asymmetric infrared abnormalities. This computer analysis is an attempt to automate the process, but it will initially only be used to aide the image reader by pointing out possible asymmetric areas. Eventually, we hope to quantitate different parameters of the infrared abnormalities in order to define the significance of the different abnormalities better, partly in terms of their presence and partly to allow for the assignment of a numerical value to each abnormality in an attempt to create an infrared abnormality index by adding together the numerical results. None of this computer-assisted image analysis can be done with the images from contact thermography, as the images from contact thermography cannot be easily digitized for image analysis. Reevaluation of the authors' data with the chi-square test, comparing the three levels of thermographic abnormality (normal, equivocal, and abnormal) to the levels of previously demonstrated prognostic indicators, showed that there were significant associations of thermographic results with a patient's age (P = 0.0113), tumor size (P = 0.0078), lymph node involvement (P = 0.0023), stage (P = 0.0001), histologic grade (P = 0.0028), and estrogen receptor status (P = 0.0429), but there was only a trend toward association with ultrasound vascularity (P = 0.0663) and no association with Ki67 (P = 0.2898). It seems that if thermographic abnormalities are so strongly associated with so many known prognostic indicators, then it is unlikely that thermographic results of the breasts would not have prognostic significance and not be a predictor of disease free and overall survival. The number of patients that were actually followed in the study to determine 5-year disease free and overall survival was not given in the article, but infrared imaging results were very significantly correlated with 5-year disease free survival (P = 0.0009) as well as tumor size (P = 0.0001), lymph node status (P = 0.0001), stage (P = 0.0001), and grade (P = 0.0001). Overall survival was not significantly related to thermographic abnormality but again was strongly associated with tumor size, lymph node status, stage, and grade. However, disease free and overall survival were quite different between the groups of breast carcinoma patients with normal and abnormal infrared images of the breast, with differences in disease free and overall survival of 21% and 9%, respectively. Establishing a significant difference in survival may require a larger follow-up study, with more patients enrolled and followed for 5 years. It may even be necessary to follow the patients for a longer period of time, as it is common for differences in disease free survival to predate significant differences in overall survival, and thus significant differences in overall survival are often only reached when the study groups are followed for a few extra years beyond the point at which significant differences are found in disease free survival. The first analysis suggesting that the results from breast thermography were related to the growth rate of breast carcinomas was reported in a paper by Gautherie and Gros.5 That analysis clearly showed that tumors classified as T1 and T2 (with diameters between 0.9 and 3.7 cm) had greater heat production than normal breast tissue and that the heat production correlated with the growth rate of the tumor. We have recently provided further evidence2 that the thermographic results are related to the growth rate of a breast carcinoma by showing that the result of three growth rate-related prognostic indicators (biochemical assay of tumor ferritin concentration, immunocytochemical analysis of Ki67, and determination of S-phase/growth fraction by flow cytometry) are strongly associated with thermographic results. Sterns et al.1 compared one of these prognostic indicators, Ki67, to results from contact thermography and concluded that thermographic results are not related to the growth rate of breast tumors. However, they did not attempt to make an actual clinical determination of growth rate with serial measurements, nor did they semiquantitate or quantitate the two other prognostic indicators that have been related to thermographic results from telethermography. Additional studies will be needed either to confirm or disprove a relationship between the results of breast thermography and survival and/or the growth rate of breast carcinomas. These studies must attempt to repeat the original studies and not rely on outdated technology, such as contact thermography, that was not even state of the art at the time of the original publications,1-5 let alone today. Future investigators should not do any more studies on single growth-related prognostic indicators, as this will only further confuse the issue. It would be best to undertake large prospective studies that look at thermographic abnormalities determined with state-of-the-art telethermography and then compare the results to the growth rate of tumors, determined by direct measurement of doubling time, and growth rate-related prognostic indicators (tumor ferritin concentration, S-phase determined by flow cytometry, and growth fraction of the tumor determined by Ki67 immunocytochemical analysis). Jonathan F. Head Ph.D., Robert L. Elliott M.D. Ph.D.