Methods for identifying, diagnosing, and predicting survival of lymphomas
View Patent ↗Gene expression data provides a basis for more accurate identification and diagnosis of lymphoproliferative disorders. In addition, gene expression data can be used to develop more accurate predictors of survival. The present invention discloses methods for identifying, diagnosing, and predicting survival in a lymphoma or lymphoproliferative disorder on the basis of gene expression patterns. The invention discloses a novel microarray, the Lymph Dx microarray, for obtaining gene expression data from a lymphoma sample. The invention also discloses a variety of methods for utilizing lymphoma gene expression data to determine the identity of a particular lymphoma and to predict survival in a subject diagnosed with a particular lymphoma. This information will be useful in developing the therapeutic approach to be used with a particular subject.
1. A method of identifying a lymphoma sample as Burkitt lymphoma (BL), the method comprising:
a) isolating gene expression product from lymphoma sample X, wherein the gene expression product is mRNA and sample X is from a patient;
b) creating one or more lymphoma type pairs, wherein each lymphoma type pair represents a combination of a first lymphoma type and a second lymphoma type, and wherein said first lymphoma type is BL;
c) for each lymphoma type pair, obtaining gene expression data for a set of genes G in said first lymphoma type and said second lymphoma type, wherein the gene expression data is mRNA expression data;
d) calculating a series of scale factors, wherein each scale factor represents a difference in gene expression between said first lymphoma type and said second lymphoma type for one of the genes belonging to said set of genes G;
e) dividing said set of genes G into two gene subsets G 1 and G 2 , wherein G 1 consists of c-myc and c-myc target genes and G 2 consists of z genes from said set of genes G with the largest scale factors, and wherein G 2 does not include any genes falling into G 1 ;
f) generating a first and second series of linear predictor scores for a set of known samples belonging to said first lymphoma type and a set of known samples belonging to said second lymphoma type, wherein said first series of linear predictor scores are based on expression of said subset of genes G 1 and said second series of linear predictor scores are based on expression of said subset of genes G 2 ;
g) contacting the gene expression product from sample X to a plurality of probes to thereby obtain gene expression data for said subset of genes G 1 and said subset of genes G 2 in sample X;
h) generating a first linear predictor score for sample X based on the expression of the genes in said subset of genes G 1 and a second linear predictor score for sample X based on the expression of the genes in said subset of genes G 2 ;
i) calculating a first probability q 1 that sample X belongs to said first lymphoma type by:
q
1
=
ϕ
(
LPS
(
X
)
1
;
μ
^
1
,
σ
^
1
)
ϕ
(
LPS
(
X
)
1
;
μ
^
1
,
σ
^
1
)
+
ϕ
(
LPS
(
X
)
1
;
μ
^
2
,
σ
^
2
)
wherein LPS(X) 1 is the first linear predictor score for sample X, {circumflex over (μ)} 1 , and {circumflex over (σ)} 1 are the mean and variance of the linear predictor scores for said set of known samples belonging to said first lymphoma type, and {circumflex over (μ)} 2 and {circumflex over (σ)} 2 are the mean and variance of the linear predictor scores for said known samples belonging to said second lymphoma type and φ(LPS(X) 1 ; {circumflex over (μ)} 1 ,{circumflex over (σ)} 1 ) is the normal density function evaluated at LPS(X) 1 with mean {circumflex over (μ)} 1 and variance {circumflex over (σ)} 1 , and φ(LPS(X) 1 ; {circumflex over (μ)} 2 , {circumflex over (σ)} 2 ), is the normal density function evaluated at LPS(X) 1 with mean {circumflex over (μ)} 2 and variance {circumflex over (σ)} 2 ; and
j) calculating a second probability q 2 that sample X belongs to said first lymphoma type by:
q
2
=
ϕ
(
LPS
(
X
)
2
;
μ
^
1
,
σ
^
1
)
ϕ
(
LPS
(
X
)
2
;
μ
^
1
,
σ
^
1
)
+
ϕ
(
LPS
(
X
)
2
;
μ
^
2
,
σ
^
2
)
wherein LPS(X) 2 is the second linear predictor score for sample X, {circumflex over (μ)} 1 and {circumflex over (σ)} 1 are the mean and variance of the linear predictor scores for said set of known samples belonging to said first lymphoma type, and {circumflex over (μ)} 2 and {circumflex over (σ)} 2 are the mean and variance of the linear predictor scores for said known samples belonging to said second lymphoma type and φ(LPS(X) 2 ; {circumflex over (μ)} 1 ,{circumflex over (σ)} 1 ) is the normal density function evaluated at LPS(X) 2 with mean {circumflex over (μ)} 1 and variance {circumflex over (σ)} 1 , and φ(LPS(X) 2 ;{circumflex over (μ)} 2 , {circumflex over (σ)} 2 ) is the normal density function evaluated at LPS(X) 2 with mean {circumflex over (μ)} 2 and variance {circumflex over (σ)} 2 ;
wherein sample X is classified as BL if q1 and q2 are both greater than 90%.
2. The method of claim 1 , wherein z is 100.
3. The method of claim 1 , wherein said second lymphoma subtype is DLBCL.
4. The method of claim 3 , wherein said second lymphoma subtype is selected from the group consisting of ABC, GCB, and PMBL.
5. The method of claim 1 , wherein said scale factors are t-statistics.
6. The method of claim 1 , wherein the linear predictor scores are calculated by:
LPS
(
S
)
=
∑
j
t
j
S
j
,
wherein S j is the expression of gene j in a sample S and t j is the scale factor representing the difference in expression of gene j between a first lymphoma type and a second lymphoma type.