Infections are the most important of all fatal com­plications in the first year after kidney transplant. They often differ in the severity of their course and can manifest in atypical symptoms. Further, their complex presentation significantly impedes diagnosis and treatment selection. Tuberculosis is unique among infections that affect patients posttransplant because it is accompanied by significant difficulties in detection, treatment, and prevention. The clinical application of a decision support system that can predict the likelihood (percentage) of patients developing posttransplant tuberculosis appears promising.

Key words : Anti-TB therapy, Immunosuppressive therapy, Renal transplant

Introduction

Kidney transplant is the treatment of choice for patients with end-stage renal disease. This type of renal replacement therapy provides the greatest survival time, maintains quality of life, and allows patients to reach maximum levels of socioeconomic rehabilitation. However, at this stage in clinical transplant, several unresolved problems remain, one of which is the control of communicable com­plications, that is, infections.

Infections are the most common and important of all fatal complications during the first year after solid-organ transplant. They often differ in the severity of their course and can manifest in atypical symptoms. In addition, their complex presentation can sig­nificantly impede diagnosis and treatment selection. Tuberculosis (TB) is unique among infections that affect patients posttransplant. This disease is characterized by significant difficulties in detection, treatment, and prevention.

The presence of a graft and drug-induced immunosuppression in the patient create difficulties in identifying and treating all serious comp­lications, including infections such as TB that occur post­transplant. These factors mask the clinical presentation of certain conditions, which may be TB-related, suppressing the development of pathog­nomonic symptoms and in turn significantly affecting patient outcomes. The use of modern diagnostic and computational tools to facilitate decision making can help clarify case management in these patients.

The aim of this study was to investigate existing difficulties in posttransplant TB diagnosis and treatment and to develop possible solutions. To achieve these aims, we formulated the following tasks: Designate factors that determine the diagnosis and pathophysiology of posttransplant TB; Create a TB detection tool for patients and assess its effectiveness; Suggest ways to improve the diagnosis and treatment of TB after transplant using tools to assist decision-makers.

Materials and Methods

This study was performed in 2 locations: the Samara Center of Organ and Tissue Transplantation (SCO&TT), Samara, Russia, and a scientific research center—the Organ Transplant Program at the Boris Petrovsky Scientific Center of Surgery, Russian Academy of Medical Sciences, Moscow. There were 2 groups of patients: group 1 included 610 patients who underwent transplant at SCO&TT, and group 2 included 449 patients at the Petrovsky Scientific Center of Surgery who were infected with TB. We analyzed all numerical data contained in the patients’ electronic health records (including anthropometric data; medical history; and clinical, laboratory, and instrumental data). Data analysis was performed using the cluster, discriminatory, and multifactor analysis system and by constructing a self-organizing Kohonen neural network devised at the breakthrough research center IT-Medicine, based on Samara State Medical University, Samara, Russian Federation). Statistical analyses were performed using software (SPSS version 11.0, SPSS Inc., Chicago, IL, USA).

Results

Posttransplant TB was identified in 24 patients (3.9% of 610 investigated patients). The patients that we investigated were categorized into several age groups. Patients’ age varied from 16 to 62 years. The greatest number of TB cases (n = 17) occurred in patients aged 25 to 34 years.

The peak in TB incidence was recorded at month 6 after kidney transplant as well as in the late postoperative period (5-10 y after transplant). We hypothesize that the first wave of disease incidence was associated with reactivation of the patient’s infection, and the second wave was associated with the process of reinfection.

The methods used to identify TB in the patients studied did not fit into existing approaches. Thus, half of the TB-infected patients were identified using radiologic methods. The percentage of TB cases detected postmortem was very high. For the most part, these were cases whose management had been neglected due to improper process or prolonged diagnostic timelines.

Weakness and fever were the most frequent symptoms of the TB disease course. Most of the patients had respiratory symptoms, as well as weight loss and night sweats. Nausea, vomiting, and gastrointestinal bleeding were seen in patients with extrapulmonary forms of TB. The high rate of diagnostic errors was based on the fact that the symptoms were nonspecific.

Tuberculosis was detected significantly more frequently in patients with comorbid diseases such as Cytomegalovirus and BK polyoma virus infection. Likely, the development of these infections indicated hyperimmunosuppression in these patients—a pathogenic factor contributing to posttreatment TB.

It is possible to devise approximately 200 treatment regimens from combinations of existing immuno­suppressive drugs. The most frequently used immuno­suppressive drug regimen in these patients was a calcineurin inhibitor + a mammalian target of rapamycin inhibitor + azathioprine/mycophenolate acid (mycophenolate mofetil) + glucocorticoids. No significant correlation was found when comparing immunosuppressive therapy regimens and the frequency of TB after transplant; there were several components of anti-TB therapy, pharmacological groups of immunosuppression medicine, and anti-TB therapy (ATT) and others.

In addition to immunosuppressive therapy, most patients were receiving multicomponent ATT. This treatment varied depending on the sensitivity of the identified Mycobacterium tuberculosis, indicating the importance of laboratory research. The ef­fectiveness of treatment regimens in patients also differed. This was related to the timing of establishing a correct diagnosis. However, patients were able to achieve favorable outcomes through timely initiation of the multicomponent ATT in conjunction with low maintenance doses of immunosuppressive drugs. Mortality rate in the group treated with ATT was significantly lower (5; 20.8%) than in patients who did not receive ATT (19; 79.2%). The number of therapy components was of great importance. Thus, graft loss or death of the patient was observed in 24 of cases (100%) with only a 1- or 2-component treatment.

Because of the heterogeneity of the data and the lack of specific pathognomonic symptoms, it was not possible to create a model for identifying post­transplant TB as a probability calculator. However, it was found that the most effective means of predicting TB after transplant is the use of an artificial neuronal network. This network is implemented based on the hybrid computing cluster “Veterok,” developed at the Siberian State Medical University (SSMU, Tomsk, Russia) (NVIDIA Tesla/Xeon Phi), whose peak performance is 14 TFLOPs (an acronym for FLoating-point Operations Per Second). This measure of computer performance), is based on center IT-Medicine (Samara, Medical University, Samara, Russia) This network was tested using the medical records of 546 patients and enabled the prediction of the probability (percentage) of the patients developing posttransplant TB.

The system is integrated into the automated information system “Transplantation” and comprises a clinical decision support system. It continuously scans electronic medical records and gives a warning in cases in which there is an increased risk of complications. Thus far, its use has accurately detected TB in 3 patients and suspected TB in 5 patients, which are now under medical control in SCO&TT. Using this system in transplant is expected to improve the treatment results of patients. It also promises to add to the data analyzed in the pre existing system of automatically recognized diagnostic images.

Conclusions

The factors that predict a high probability of a posttransplant TB diagnosis include typical history, symptoms (fever of unknown origin, local symptoms, which are explained at extrapulmonary TB), ad­ditional studies (which include all laboratory tests to reveal TB), and immunosuppressive therapy, without the presence of pathognomonic symptoms. However, these factors are valuable only in an integrated context.

Posttransplant TB can be characterized as a new nosologic form because it has several epidemiologic, pathophysiologic, and clinical features as well as different treatment protocols and outcomes. A clinical decision-making system can be an effective means for detecting TB with high sensitivity and reliability. To improve the treatment of TB after transplant, clinicians need to view this pathology as a new nosologic form, maintain high alertness to its presence, and make the transition from a symptomatic approach to a more integrated diagnostic approach through the use of clinical decision-support systems.