Telepathology is the practice of pathology at a distance, transmitting images using telecommunication methods for second opinion and/or diagnostic assistance, or for educational purposes. It may be the only means of consultation for some pathologists.
To retrospectively review and evaluate a subset of telepathology consultations from June 1996 to March 1997, and to determine the concordance between the telepathology diagnosis of the contributor and pathologists at the Armed Forces Institute of Pathology (AFIP), Washington, District of Columbia, as well as the concordance between the telepathology diagnosis and the glass slide diagnosis, when available.
Photocopies of de-identified telepathology reports from the AFIP during a 15-month period between June 1996 and March 1997 were reviewed. Contributor versus telepathology diagnosis was graded as 1 (complete agreement), 2 (partial agreement), 3 (disagreement; usually a diagnosis of benign versus malignant), and deferred. Data were analyzed using descriptive statistical methods.
Of the 262 cases, 194 (74%) were in complete agreement with the contributor's diagnosis, 34 of 262 (13%) were in minor disagreement, and 21 of 262 (8%) were in major disagreement. Diagnoses were deferred in 5% (13 of 262) of cases.
Using commercial off-the-shelf technology and despite telecommunication challenges during that time, the AFIP demonstrated that telepathology could be conducted reliably.
Telepathology is the practice of pathology at a distance, transmitting images using telecommunication methods for second opinion and/or diagnostic assistance, or for educational purposes.1,2 The receiving pathologist reviews selected images digitized by the remote pathologist and gives an interpretation. It is often the only means of consultation or second opinion for many single-practice pathologists or pathologists worldwide who practice in remote locations.
The Armed Forces Institute of Pathology (AFIP) started in 1862 as the Army Medical Museum “to collect and catalogue the specimens that had been accumulating from men fighting in the American Civil War.”3 The AFIP's mission was 3-fold—consultation, education, and research—but it was the consultation mission that was considered by many pathologists to be its most important role.3 In 2007 alone, more than 50 000 cases were processed from pathologists throughout the world. This was achieved by the AFIP's 22 departments and branches with a staff of more than 120 pathologists, many of whom were renowned experts in their specialties.
The telepathology program was launched in 1993 to provide expert consultation to US pathologists and overseas pathologists, especially pathologists in solo practice who did not have easy access to experts.4,5 The telepathology unit was modeled according to standard AFIP procedures. Although initially only proprietary telepathology systems were used for digitizing images, an open, commercial off-the-shelf image-capturing system was started in 1995. Although videoconferences were regularly held with the National Cancer Center in Japan to discuss educational cases, the program was mainly a store-and-forward system. Despite the lack of whole slide imaging systems at the time, however, it allowed any pathologist with access to the Internet and a high-resolution camera to send teleconsultations to the AFIP.4 The images sent were static, partial images of the specimens, selected by the submitting pathologist rather than images scanned by whole slide imaging. Even today, when whole slide imaging technology is quite common, this is not available in many of these locations. During the period of review, the AFIP telepathology network received consultations from approximately 19 hospitals and medical centers worldwide.
The images, often compressed by JPEG format, were sent as email attachments or via FTP to the telepathology service. Although the JPEG format can be compressed in a “lossless” format, most users submitted JPEG images using a “lossy” compression algorithm, which was the default mode at that time. While this yielded smaller file sizes and faster transmission, some photographic information would be lost. However, this loss of information did not likely adversely affect the quality of the images, nor did it have a negative impact on the accuracy and confidence level of diagnosis.6 Upon receipt of images, cases were assigned an AFIP accession number, and the pertinent AFIP department was immediately informed. The assigned pathologists either came to the telepathology laboratories to review the consultations on a large high-resolution monitor or viewed the images in their own departments through an Intranet. The reports were dictated by the consultant pathologists, transcribed, and sent by fax and postal service. The referring pathologists were called by telephone in urgent cases. In most cases, reports were communicated to the referring pathologist within 4 hours after receiving the cases. In challenging cases, glass slides and paraffin blocks were requested for follow-up studies.
In 2011, the AFIP was “disestablished” together with the Walter Reed Army Medical Center in Washington, DC, in accordance with the Base Realignment and Closure Act.3 Its massive, one-of-a-kind tissue repository of about 55 million glass slides, 31 million paraffin blocks, and more than 500 000 wet tissue samples, which some have described as an “international treasury,” is now under the control of the Joint Pathology Center.3 The Joint Pathology Center has assumed many of the consultative functions of the AFIP.
The purpose of the current study was to retrospectively review and evaluate a subset of telepathology consultations from June 1996 to July 1997 in an expert institution serving military hospitals, Veterans Administration medical centers, and academic and civilian pathologists worldwide. The aim was to determine the concordance between the telepathology diagnosis of the sending facility and AFIP pathologists, and the concordance between the telepathology diagnosis and the glass slide diagnosis, when available.
MATERIALS AND METHODS
The data source for this review was photocopies of de-identified telepathology reports during a 15-month period between 1996 and 1997. No personally identifiable information was included in the reports. Patients were not identifiable by the investigators or others, either directly or indirectly. With the closure of the AFIP, linkage to patient information is not possible. The research team had no direct patient care responsibilities with the patients or referring pathologists.
Data were entered into an Excel (Microsoft Corp, Redmond, Washington) spreadsheet and were tabulated under the following categories: AFIP department, contributor diagnosis, and de-identified clinical history, when available; AFIP telepathology diagnosis; AFIP glass slide diagnosis (when available); additional comments; and number and quality of images.
Contributor versus telepathology diagnosis was graded as 1 (complete agreement), 2 (partial agreement), and 3 (disagreement; usually a diagnosis of benign versus malignant). In some instances, diagnosis could not be rendered because of image quality, which was graded as good or poor. In some cases, glass slides and paraffin blocks were requested, which were then reviewed and compared with the contributor's diagnosis and the telepathology diagnosis. A category was also created for cases where diagnosis was deferred. Data were analyzed using descriptive statistical methods.
The National Institutes of Health Office of Human Subjects Research designated this research as exempt from Institutional Review Board review.
RESULTS
In this study, a total of 262 reports during a period of 15 months in the years 1996 to 1997 were reviewed. Table 1 shows the distribution of cases by AFIP department. The departments of dermatopathology; gynecologic and breast; and hepatic and gastrointestinal pathology had the 3 highest number of referrals. The least number of consults were in the departments of ophthalmic; infectious and geographic; and veterinary pathology. In most cases there was a note about the number and quality of images sent. The average number of images was 10 (range, 1–32; SD = 5.4). Images were graded as being of either good or poor quality.
Of the 262 cases, 194 (74%) were in complete agreement with the contributor's diagnosis, 34 of 262 (13%) were in minor disagreement, and 21 of 262 (8%) were in major disagreement (Figure). In 13 of 262 cases (5%), diagnoses were deferred.
The deferred cases included mostly gynecology and breast pathology, hematopathology, and dermatopathology cases (Table 2). Of the 13 deferrals, 54% (7 of 13) were confirmed by reviewing glass slides. All of these 7 cases (100%) matched with the contributor's diagnosis. In the remaining 6 cases, glass slides were unavailable for review.
In the cases that had major disagreement (21 of 262; 8%), the major reason was rendering a malignant versus a benign diagnosis (Table 3). The contributor's diagnosis in 9 of 21 cases (43%) was a malignancy, whereas AFIP Telepathology reported them as benign entities. Only 1 of these 9 was reverted back to malignancy upon glass slide review. A total of 3 of 21 cases (14%) were called benign by the contributor but were diagnosed as malignant by the AFIP.
DISCUSSION
This study focused on telepathology data from a specific time period at a single institution to compare with the contributor's diagnosis and diagnosis based on slides, if they were available, to confirm and concur with the telepathology diagnosis.
The technical challenges that faced the contributing pathologists in remote locations were substantial: slow and unreliable Internet connections through dial-up modems (theoretical transfer speed of 56 kilobits per second, but only 40–50 kilobits per second or less usually) to an Internet service provider, which often resulted in dropped connections, and the lack of high-resolution cameras.4,5,7 Many of these laboratories also lacked access to immunohistochemical stains to resolve challenging cases.
Similar studies of this kind have been conducted either in single institutions or with international collaboration. A study from China in 2011–2013 showed that the concordance rate was 59.8%.8 In a similar study at the University of Pittsburgh (Pittsburgh, Pennsylvania) in collaboration with China, authors reported a concordance rate of 25.6% among the 54.7% of cases where an outside diagnosis was provided.9 Another study in India examined the role of telepathology in rural institutions and found a concordance rate of 90.2%, with approximately 9.7% being marked as major discrepancies.10 Our review showed a concordance rate of 74%, which was well within the range of the above published observation. In a 12-year study based on teleconsultations between Iron Mountain Medical Center (Iron Mountain, Wisconsin) and the Milwaukee Department of Veterans Affairs Medical Center (Milwaukee, Wisconsin) using a robotic surgical telepathology system, the discordance rates were as low as 0.12% to 0.77%, whereas deferral rates were 2.5% to 28.7%.11 Our study found a discordance rate of 8% (21 of 262) and a deferral rate of 5% (13 of 262). Most discordant cases were followed up by glass slides and reached concordance. Of 13 deferred cases, 5 were revisited on glass slides sent by a contributor, and they all matched the contributor's diagnosis (100% concordance; Table 2). A possible explanation for the high level of discordance, especially the overinterpretation of malignancies by the contributing pathologists (9 of 21; 42%), is that cases referred to the AFIP were likely more difficult than routine cases. Many of the consultations submitted were cases to resolve questions of whether lesions were benign versus malignant. As a world center for pathology consultations, the AFIP routinely received challenging and really difficult cases from pathologists worldwide. Moreover, there are also other contributing factors that may affect the concordance/discordance rates of telepathology as in glass slide consultations: difficulty of cases; training and expertise of the submitting pathologists; and laboratory capabilities, such as immunochemistry and special stains, etc.
In this study, the disease or department distribution was heavily biased toward dermatopathology, followed by gynecologic and breast and hepatic and gastrointestinal pathology (Table 1). We posit that this pattern results from the difficulty in cases within these pathologic specialties. In previously published reports, similar trends were observed, where most consults were for the gynecologic and gastrointestinal departments.6,7 In the study from China, the top 3 consultations were for the digestive system (17.3%), the gynecologic system (16.7%), and head and neck (15.7%).8
In traditional pathology consultation services (ie, nontelepathology), the major discrepancy in diagnoses ranges from 10% to 64%, worldwide.11–13 A study of pediatric neoplasms at St Jude Children's Research Hospital (Memphis, Tennessee) from referring international sites showed that the major disagreement rate between international sites and St Jude's was higher than that among US institutions (13% to 37% among the countries studied).13 The authors attributed this variation to the lack of availability of immunohistochemical stains and pathologists' training in the diagnosis of pediatric malignancies. Another study, using whole slide imaging, found major discrepancy rates as low as 1.4%, although the report retroactively scanned frozen section slides of intradepartmental consultations, as opposed to “live” cases shared between institutions.14,15 The discrepancy rates of the AFIP telepathology services (21 of 262; 8%) suggest that the telepathology diagnostic accuracy level is near that of traditional pathology consultation services, if not its equal. Moreover, there is often wide variability in concordance and discordance rates based on histologic type, hence the need for a second opinion, as is often done in standard glass slide consultations and in telepathology as well.
The telepathology service served as a “second opinion” source for challenging cases. Every report was transmitted with the statement that “the report is not valid unless countersigned by the contributing pathologist.” The importance of this second opinion is immense and has been proven to change therapy. For example, in a study of 855 radical prostatectomies performed at Johns Hopkins Hospital (Baltimore, Maryland), a retrospective review of the pathology reports confirmed cancer in 844 cases (98.8%) by needle biopsy.12 Of the remaining 11 cases, 9 (1%) were considered atypical and 2 (0.2%) were benign, indicating that some cases may benefit from a second opinion.
Internationally, telepathology is driven by the need to improve access to specialist pathologists in areas where they are not available. A successful collaboration leads to multiple benefits—not only does it ensure optimal patient care, which is the clinical team's primary responsibility, but it also contributes to effective communication and serves as a learning resource. There are factors that need to be present to make this effort successful, and these include: addressing legal and regulatory issues, cost, cultural factors, equipment, information technology infrastructure, and leadership support.7 Diagnostic differences in naming certain pathologic specimens also needs to be clarified between the referring site and the expert site. Within departments, there are issues that need to be improved for running an optimal service unit, such as, local information technology support, confidentiality, Health Insurance Portability and Accountability Act issues, and language barriers.8 In spite of the challenges, the AFIP telepathology program demonstrated more than 20 years ago that this could be accomplished successfully in a manner that benefited everyone.
This research was supported by the Intramural Research Program of the National Institutes of Health, National Library of Medicine, and Lister Hill National Center for Biomedical Communications.
References
Author notes
From the Lister Hill National Center for Biomedical Communications, National Library of Medicine, National Institutes of Health, Bethesda, Maryland.
The authors have no relevant financial interest in the products or companies described in this article.
