### Description

A mathematical model can be used to estimate the expected survival for a patient with Hodgkin's disease at the time of diagnosis. This can help tailor the treatment to meet the needs of the patient.

Patient population:

(1) 5,023 patients in the International Database on Hodgkin's Disease (IDHD).

(2) Most patients were treated with randomized trials or standard protocols.

Data analysis: The log-normal model gave the best fit for the data.

Parameters:

(1) disease stage

(2) age of the patient in years

(3) histologic type

(4) systemic symptoms

(5) serum albumin

(6) distribution of disease involvement

(7) patient gender

 Parameter Finding Value Stage I present 1 absent 0 Stage II present 1 absent 0 Stage III present 1 absent 0 age in years age histologic type lymphocyte predominant 1 nodular sclerosing 1 mixed cellularity 1 lymphocyte depleted 0 systemic symptoms A symptoms 1 B symptoms 0 serum albumin percent (from 0 to 100) of frequency distribution for center involved area distribution <= 3 involved areas above diaphragm 1 all other distributions (see below) 0 gender female 1 male 0

where:

• Stage IV disease would have Stages I, II and III being absent

• B symptoms indicate one or more of the following: night sweats, fever > 38°C or weight loss > 10% in the past 6 months

• other distributions include (1) >= 4 supradiaphragmatic areas, (2) any subdiaphragmatic area, (3) areas on both sides of the diaphragm

• serum albumin is expressed as "the percentile of the frequency distribution within each center"; this was not further defined. This was necessitated by variation in analytic methods used and reference ranges between medical centers. However, this could either represent (a) percentile for distribution of all patients seen at the center, or (b) percent of the normal distribution. The data for (a) is often not known and would be affected by the population mix. The data for (b) would essentially give a value of < 5 if below the normal reference range used (I would assume 0 would be entered then for very low values).

median expected survival in months =

= EXP(3.75 + (1.25 * (value Stage I)) + (0.77 * (value Stage II)) + (0.46 * (value Stage III)) - (0.00046 * ((value age)^2)) + (0.85 * (value histologic type)) + (0.42 * (value systemic symptoms)) + (0.24 * (LN(value serum albumin))) + (0.25 * (value for involved area distribution)) + (0.25 * (value for gender)))

where:

• EXP(X) = e^(X)

probability that a patient will survive at least X months =

= 1 - (Gaussian distribution function ((LN(X)) - (LN(median expected survival in months))))

where:

• (LN(median expected survival in months)) is the value of the equation above within the "EXP()"

• The Gaussian distribution function is not included with Excel.

The patient could also be assigned to one of 5 prognostic classes which ranged from excellent (class 1) to poor (class 5):

 Class Median Expected Survival in Months 1 >= 723 2 377 - 722 3 178 - 376 4 64 - 177 5 < 64 months

(from pages 248-249).