Second Oncology Example

Second Oncology Example

Sample Size for non-inferiority for survival studies

Many cancer drugs although effective in prolonging survival put patients at considerable risk for serious adverse events. A new drug with an improved safety profile would be of interest even if it did not increase survival as long as survival was not decreased. For example, suppose the standard drug results in a median survival of 12 months. The new drug might be considered to be not inferior as long as the median survival was not lower than 10 months. The study will be designed as a non-inferiority trial with a survival outcome.

Start by choosing the survival table for two groups for equivalence. The sample size table for Two group test of equal exponential survival appears.

Entering the values

This table calls for specification of the exponential survival parameters. Since we only know the values for median survival, we select the Assistants menu Conversion Between Parameters option.

In the STT00 Conversion between Parameters side table, enter 10 for Group 1 median survival and 12 for Group 2 median survival. The table automatically calculates the hazard ratio.

We find that these values correspond to a hazard ratio no larger than 1.2 and that the exponential parameters for survival times expressed in months are 0.0693 and 0.0578. Using Copy and Paste Unformatted, we transfer the exponential parameter for the standard treatment and the hazard ratio to STE0, the survival non-inferiority table.

We then enter information about accrual period, followup period, and expected dropout rate, being sure to specify these in terms of months. For this example, we assume that the accrual period will be 18 months and that the maximum length of followup will be 24 months. In other words, those entered in the first month of the trial will be followed for 24 months and those entered at 18 months after the start of the trial will be followed only for 6 months.

The required total number of events is 744. The number of subjects who must be entered into the trial to achieve the number of events under the alternative hypothesis that both exponential parameters are equal to that specified for the standard is 664 per group (1328 total). If the null hypothesis were true and the hazard ratio were 1.2, the required number of events might be achieved with fewer subjects (see the last row in the table).

To get a statement for this result, click on the Statement icon. The statement will appear in a separate dialog box. You can print that statement or click on the Store button to store the statement with the STE0 table.

Adding additional columns and formatting the table

To request a different power, we use the second column of the table. Copy everything in the first column, paste into the second column and enter 90 in the Power row for the second column.

To name the columns, choose the Edit menu Edit Column Names option. Enter 80% Power for Column 1 and 90% Power for Column 2. Click on the OK button to update the column names in STE0 with the new names.

To display more decimal digits in the row for the standard exponential parameter, right click in that row, and select Format decimal displays for selected rows, and enter 6 in the dialog box.

Plotting sample size information

To get a plot relating the sample size to the specified hazard ratio, select the columns for plotting, then choose the Plot menu Plot User-selected rows. Define the X-axis and the Y-axis, then set the X-axis range.

This plot shows that using a non-inferiority limit of 1.3 would require less than half the number of patients. However, such a non-inferiority limit would imply that a median survival for the new drug potentially as low as 9.2 months would be acceptable.

You can print plots, copy to the clipboard for pasting into documents, or save to a file.

Printing, saving, and pasting results into documents

You can print the sample size tables (together with notes, statements, and references) or save to a file (together with notes and statements). You can also copy sample size tables, statements, notes, references, and guidecards to the clipboard and paste them into your document. The Greek symbols will appear in the symbol font, while the rest of the text will appear in the font of your choice.

We show the pasted table on the right.