Breast cancer therapy planning – a novel support concept for a sequential decision making problem

Abstract

Breast cancer is the most common carcinosis with the largest number of mortalities in women. Its therapy comprises a wide spectrum of different treatment modalities a breast oncologist decides about for the individual patient case. These decisions happen according to medical guide lines, current scientific publications and experiences acquired in former cases. Clinical decision making therefore involves the time-consuming search for possible therapy options and their thorough testing for applicability to the current patient case.This research work addresses breast cancer therapy planning as a multi-criteria sequential decision making problem. The approach is based on a data model for patient cases with therapy descriptions and a mathematical notion for therapeutic relevance of medical information. This formulation allows for a novel decision support concept, which targets at eliminating observed weaknesses in clinical routine of breast cancer therapy planning.

Appendices

Appendix A: Patient case model

Exemplary status attributes as introduced in [24, 57] are:

• ACR describes the tissue density in the breast based on the ordinal value domain {I, II, III, IV}.

• Age describes a patient’s age in years.

• BIRADS assesses a breast imaging report in terms of the risk for carcinoma based on the ordinal domain {0, I, II, III, IV, V, VI}.

• Charlson Score encodes the presence and severity of comorbidities with ordinal values from {0, 1, 2, 3, 6}.

• Comorbidities describes the patient’s anamnesis with former and present diseases in terms of about 1000 values, see for example [13].

• Distant metastases provides information about the general occurrence of metastases in other body parts encoded by the M-category with the ordinal value domain {0, 1} and other attributes specifying the locations.

• Estrogen receptor describes the cell degree of interference to structural changes by hormonal influence. It consists of an integer Percentage value and ER, which describes the receptor status as + or − according to the percentage.

• Finding location describes the location of a finding in the breast in terms of the Side with the ordinal value domain {left,right} and the Clock time position with the value range {1h, 2h, …,12h}.

• Gender has the value options male and female.

• Grading quantifies the level of differentiation between normal and tumor cells with a value from {1, 2, 3}.

• Human epidermal growth factor receptor 2 describes the status of a receptor, which stimulates cell growth and blocks cell death, with a value from {+, −} derived from a Classification with the domain {0, 1+, 2+, 3+}.

• Karnofsky Index describes the physical performance status of a patient with percentage values {0, 10, 20, …, 100}.

• Medication describes the patient’s medical anamnesis in terms of the consumed Active ingredient, Dose and Duration, see for example [68].

• Menopausal status may attain one of the values premenopausal, perimenopausal and postmenopausal.

• Proliferation marker quantifies cell growth with a with a value from {+, −} derived from a Ki-67 integer percentage.

• Progesterone receptor consists of PR, which describes the receptor status with a value from {+,−} derived from an integer Percentage value.

• Regionary lymph node status consists of several single-value attributes, most prominently the Finding, which specifies the information source with the ordinal values c (medical imaging and others), p (pathologic verification), y (status after neoadjuvant therapy) and suitable combinations, the N-category with the value domain

  $$\begin{array}{@{}rcl@{}} \text{dom}(\mathit{N-category}) = \{ 0, \mathrm{0\ i+}, \mathrm{1mi}, 1, \mathrm{1a}, \mathrm{1b}, \mathrm{1c}, 2,&&\\ \mathrm{2a}, \mathrm{2b}, 3, \mathrm{3a}, \mathrm{3b}, \mathrm{3c} &\}& \end{array} $$

  whose ordinal elements basically reflect different levels of cancer occurrence in regional lymph nodes, and sn with the values + and − based on the numbers of affected and examined nodes.

• Resection distance assesses the tissue Distance given in millimeters, which was additionally removed around a tumor by surgical therapy, with the parameter R on the value domain {0, 1, 2}.

• Tumor size consists of several single-value attributes, most prominently the Finding, which specifies the information source with the ordinal values c (medical imaging and others), p (pathological verification), y (status after neoadjuvant therapy) and suitable combinations, and the T-category with the value domain

  $$\begin{array}{@{}rcl@{}} \text{dom}(\mathit{T-category}) = \{ 0, \text{is}, 1, \mathrm{1mi}, \mathrm{1a}, \mathrm{1b}, \mathrm{1c},&&\\ 2, 3, 4, \mathrm{4a}, \mathrm{4b}, \mathrm{4c}, \mathrm{4d} &\}& \end{array} $$

  whose ordinal elements reflect size classifications of the tumor mainly based on its diameter.

• Tumor type has the domain

  $$\begin{array}{@{}rcl@{}} \text{dom}(\mathit{Tumor\; type}) & = & \{ \text{ductal carcinoma in situ},\\ && \text{invasive ductal carcinoma},\\ & & \text{invasive lobular carcinoma},\\ && \text{intraductal papillary carcinoma}, {\dots} \} \end{array} $$

  with about 50 nominal values representing different medical tumor classifications.

The following exemplary attributes describe diagnostic and therapeutic steps in case history as specified in [4, 34, 35], based on the uniform value domain {discarded, recommended,scheduled,carried out}, which in case of long duration procedures also contains the value ongoing:

• AI upfront is an adjuvant endocrine therapy based on the application of an aromatase inhibitor like Anastrozole, [1], or Letrozole, [7], for a period of five years.

• AI → Tam is an adjuvant endocrine therapy, which comprises an application of an aromatase inhibitor for 2-3 years followed by an application of the selective estrogen receptor modulator Tamoxifen, [2], for another period of 2-3 years up to the therapy duration of 5 years.

• Axillary sentinel lymphonodectomy is a surgical procedure on the axillary lymph nodes.

• Biopsy of the breast means the extraction of tumor tissue for examinations.

• Breast conserving therapy is a tissue resection in the breast.

• 6 × D A C q 3 w is an actual standard chemotherapy regimen with 6 cycles of Docetaxel, Adriamycin and Cyclophosphamid.

• 4 × E C q 3 w → 4 × D o c q 3 w is an actual standard chemotherapy regimen with 4 cycles of Epirubicin and Cyclophosphamid followed by 4 cycles of Docetaxel, [33, 34].

• 4 × E C q 3 w → 12 × P a c q 1 w is an actual standard chemotherapy regimen with 4 cycles of Epirubicin and Cyclophosphamid followed by 12 cycles of Paclitaxel, [33, 34].

• Interdisciplinary postoperative tumor board is a case conference at KEM for treatment planning of patients with surgical therapies already carried out.

• Interdisciplinary preoperative conference describes a case conference at KEM where especially the surgical treatment of patients is planned.

• Inspection of the breast is a diagnostic examination of the breast.

• Mammography is a diagnostic examination of the breast by means of medical imaging.

• Mastectomy means a surgical breast removal.

• Palpation of the breast is a diagnostic examination of the breast.

• Postoperative standard radiation therapy is a treatment of the breast by means of radiation after a breast conserving surgery has been done.

• Breast sonography is a diagnostic examination of the breast by means of ultrasound imaging.

• Staging combines several diagnostic examinations by means of medical imaging.

• Tam is an adjuvant endocrine therapy, which features the application of Tamoxifen, [2].

• Tam → AI is an adjuvant endocrine therapy, which comprises an application of Tamoxifen, [2], for a period of two or three years followed by an aromatase inhibitor, [1, 7], for another period of 2-3 years up to the duration of 5 years of endocrine therapy.

Appendix B: Medical relevance model

Exemplary evaluation attributes, some of them specified in [4, 43] with the values ordered from worst to best are:

• AGO with the domain {−−, −, +/−, +, ++} provides a general preference indicator for therapies.

• Absolute contraindication assesses the presence of some attribute values in a case, which definitely prevent the application of some therapy to a case, with the values +, which means fulfillment, and -, which stands for violation.

• Indication describes the applicability of a therapy to a case based on strict value requirements on some attributes with the values + for fulfillment and - for violation.

• Matching describes the comparability of patient cases according to some attribute values based on the values + for match, and - for mismatch.

• NCCN with the domain {1,2A,2B,3} provides a general preference indicator for therapies.

• Relative contraindication assesses the presence of attribute values in a case, which possibly prevent the application of a therapy concept to a case, with the values + for fulfillment and - for violation.

• Relative indication describes the applicability of a therapy to a case based on optional value requirements on some attributes with the values + for fulfillment and - for violation.

• 5-y-DFS quantifies the fraction of patients with disease-free survival after 5 years with percent values from the domain {0, 0.1, 0.2, …, 99.8, 99.9, 100} for a specific therapy concept.

• 5-y-OS quantifies the fraction of patients with overall survival after 5 years with percentages from the domain {0, 0.1, 0.2, …, 99.8, 99.9, 100} for a specific therapy concept.