It is probably a translocation written t(9,21,22) instead of the regular CML translocation t(9,22), so ask your Onc about that. In unusual cases like yours, the Philadelphia Chromosome is a 3-way translocation, which occurs in about 5% of CML cases. The TKI drugs should work against most of these translocations, although there are some where they are less effective.
Regarding whether the PCR will work, I would first try to reaccomplish the PCR to see if the PCR failed due to a lab error or spoiled sample. If you have done a couple PCRs and both were negative, your Onc should try to switch to a lab where the PCR test uses a different control gene. Most PCRs now use ABL as the control gene in the PCR test. During the original translocation, pieces of chromosomes 9 and 22 normally both lose a piece, and each of those two pieces attach to the wrong chromosome (the piece of 22 goes to 9, and the piece of 9 goes to 22 where it creates the Philadelphia Chromosome). But in your case, I assume that during the translocation, you had three chromosomes lose a piece (9, 21, and 22). Your 22 picked up the piece of 9 and created the Philadelphia Chromosome causing the CML. Then the 21 picked up the piece of 22, and then 9 picked up the piece of 21. I am speculating a bit here, but apparently from what I could find, in your situation the 9 looks normal length during cytogenetic analysis since the piece of 21 it picked up was about the same length as the piece it lost, so the new 9 looks normal under the microscope even though it is not (cytogentics uses chromosome length as a primary indicator of normality). Most of us have two abnormal chromosomes, but you would have three (9, 21, and 22). Our chromosome 9 is a different length so it appears abnormal.
The reason for the above discussion is that a PCR which uses ABL (found on chromosome 9) as the control gene may be stymied by the way your chromosome 9 is linked to a piece of chromosome 21, which may somehow block the PCR's ability to use the ABL as the control gene. I am guessing about this, since I do not have enough info. So your Onc should try to use a lab which uses a different control gene such as GAPDH, b2m, or GUS. FISH will not be fooled by the chromosome issues I outlined.
http://www.ncbi.nlm....pubmed/14562124
Although the following is an old article, it discusses how the chromosome 9 can look normal in cases like yours, while the chromosome 22 is a Philadelphia Chromosome, although it is t(9,21,22) instead of t(9,22):
http://www.haematolo...nt/79/6/536.pdf