UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)
UvA-DARE (Digital Academic Repository)
Radiation induced lung damage
Seppenwoolde, Y.
Publication date
2002
Link to publication
Citation for published version (APA):
Seppenwoolde, Y. (2002). Radiation induced lung damage.
General rights
It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulations
If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.
Chapterr 4
Pulmonaryy function following high dose
radiotherapyy of non-small cell lung cancer
Katrienn De Jaeger, Yvette Seppenwookie, Liesbeth J Boersma, Corinnee Goedbloed, Sara H Mutter, Paul Baas, Joséé SA Befderbos, Joos V Lebesque
Pulmonaryy function following high dose
radiotherapyy of non-small cell lung cancer
Eighty-twoo patients with inoperable NSCLC were evaluated before and 3-4 months after highh dose radiotherapy <RT) using a computed tomography (CT) scan of the chest a singlee photon emission CT (SPECT) king perfusion scan and PFTs (forced expiatory volumee (FEVi) and diffusion capacity (T^coe))- The average rung perfusion was measured beforee RT arid at 3-4 months post-RT. m addition, the perfusion post-RT was predicted usingg a dose-effect relation for regional lung perfusion. The measured and predicted perfusionn reductions were calculated and their difference was defined as reperfusion. The meann dose to the lungs was weighted with the perfuston resulting in the mean perftraion-weightedd kmg dose. Tumor regression and reductions of PFTs were also computed. Tumorr regression was associated with an improvement of bom FEVt and T|_coc- The
reductionreduction of PFTs was correlated with the mean periusion-weighted rung dose and the predictedd perfusion reduction but the correlation coefficients were low. The other tested
variabless (mean lung dose, measured perfusion reduction and reperfusion) were not correlatedd with changes in PFTs. Reperfusion was associated with me overall lung perfusionn at baseline. In centraHy-focated tumors mere was a trend for an association betweenn reperfusion and tumor regression.
Introduction n
Radiotherapyy of tumors located within or around the thoracic cavity inevitably results in partial irradiationn of the surrounding normal lung tissue. Lung damage following radiotherapy has beenn reported in breast cancer (Taghian 2001), Hodgkin's lymphoma (Boersma 1995, Theuws 1999,, 2000), esophageal and lung cancer (Rubenstein 1988, Curran, Jr. 1992, Choi 1994, Markss 2000a). Radiation-induced respiratory toxicity ranges from an (often) asymptomatic impairmentt of lung function to fibrosis and radiation pneumonitis which can develop into a life-threateningg complication (McDonald 1995).
Severall authors observed that patient-related (age, smoking, baseline pulmonary function) andd treatment-related (radiation dose, volume, fractionation, chemotherapy and hormonal therapy)) factors are involved in the development of radiation pneumonitis (Jensen 1990, Emamii 1 9 9 1 , Allavena 1992, Niemierko 1993, Martel 1994, Bentzen 1996, Marks 1997, Johanssonn 1998, Gagliardi 2000, Robnett 2000, Taghian 2001). Furthermore, biological factorss like transforming growth factor-beta (TGF-B, Anscher 1 9 9 8 , 2 0 0 1 , Fu 2001), surfactant proteinss (Sasaki 2001) and circulating interieukine-6 (Chen 2001b) may be implicated in the processs of radiation pneumonitis. Several investigators have constructed and tested normal tissuetissue complication probability models based on three-dimensional (3D) dose distribution data too predict the probability of this severe condition (Lyman 1985, Rubenstein 1988, Niemierko 1993,, Jackson 1993, Martel 1994, Yorke 2001). Others found that simple dose-volume parameterss like the mean lung dose (MLD, Kwa 1998a), the percentage of lung volume receivingg more than a threshold dose of 2 0 Gy (Graham 1999) or 30 Gy (Marks 1997) solely orr in combination with biological factors like TGF-B levels (Fu 2001) can be used to predict radiationn pneumonitis.
Inn addition to the prediction of radiation pneumonitis, which is a binary type of complication, it iss clinically relevant to evaluate and predict the graded response of lung tissue to radiation.
ChapterChapter 4
Thiss response can be quantified by changes in pulmonary function teste (PFTs, Boersma 1995,, Marks 2000a). Predicting PFTs post-radiotherapy is of particular interest in patients with NSCLCC of whom the majority already has an impaired pulmonary function prior to radiotherapy.. Several authors (Choi 1985, Rubenstein 1988, Curran, Jr. 1992) have tried to estimatee this function by taking into account the percentage of perfused lung in the estimated irradiatedd volume. In general the residual lung function post-radiotherapy was better than predicted.. This inconsistency was attributed to the applied methodology, which did not considerr the full 3D-dose distribution.
Wee previously assessed changes of lung function in patients with Hodgkin's lymphoma post-radiotherapyy using full 3D SPECT lung perfusion and ventilation scans and correlated pre-treatmentt CT scans (Boersma 1994). We showed that the average reduction of local perfusion correlatedd with the reduction of PFTs. Theuws et al. (1998) extended this work and reported that,, in patients with relatively healthy lungs (breast cancer and lymphoma), the reduction of PFTss strongly correlated with the predicted perfusion reduction as well as with the MLD, which iss a pure dose parameter.
Forr patients with lung cancer, it is however more intricate to estimate the amount of functional lungg damage. The majority of these patients suffer from pre-existing lung disease, which is frequentlyfrequently associated with inhomogeneous lung perfusion and fluctuating pulmonary function. Furthermore,, tumor progression can contribute to functional damage. Conversely, shrinkage
off central lung tumors initially obstructing the blood flow through pulmonary vessels can lead too perfusion recovery and hence compensate for radiation-induced injury (Marks 2000b). tt has alsoo been reported that lung cancer patients may experience an improvement of their PFTs followingg radiotherapy (Choi 1994).
Seppenwootdee et al. (2000) have correlated perfusion changes on SPECT scans and 30-dose distributionss in patients with NSCLC and derived a dose-effect relation. They compared the reductionn of local perfusion as predicted based on this dose-effect relation with the actually measuredd perfusion loss as assessed from the follow-up SPECT. Eighteen of 25 evaluated NSCLCC cases experienced a perfusion loss, which was on average 7.2% less than predicted andd this was ascribed to local reperfusion.
Becausee of the parallel structure of the lung (Yorke 1993), it can be hypothesized that the sum off changes in regional perfusion may correlate with the overall lung function as measured by PFTss (Fan 2001a, 2001b). Furthermore, lung perfusion is often inhomogeneous in NSCLC patients,, so that it is conceivable to assume that the radiation dose delivered to non-perfused regionss contributes less to functional lung damage.
Therefore,, the main objective of this study was to evaluate radiation-induced changes in overalll pulmonary function (as measured by PFTs) in relation to changes in perfusion, dose andd perfusion-weighted dose parameters. Following the observation that PFTs can improve in NSCLCC patiënte after radiotherapy and the occurrence of reperfusion, we further examined whetherr reperfusion (recovery of functional (perfusion) damage) translates into an improvementt of PFTs. In addition, the impact of tumor regression on reperfusion and PFTs was studied. .
PulmonaryPulmonary function folkmng high dose radiotherapy
Patientss and methods
Patient,, tumor and treatment characteristics
Eighty-twoo patients with medically inoperable or locally advanced NSCLC and good prognostic factorss (weight loss < 10%, Eastern Cooperative Oncology Group performance status less thann 2) referred to the department for radical RT were included in this study (Table 1). Eligibility criteriaa were: presence of visible tumor on a diagnostic chest CT scan, availability of CT and SPECTT scans prior to irradiation and at 3-4 months follow-up (all acquired in RT treatment position,, which was supine with the arms raised above the head in a forearm support), baselinee PFTs (see below) with at least a FEVj and/or TLCOc at 3-4 months follow-up. Patients withh disease progression were excluded from the study to avoid the confounding effect of tumorr progression on the evaluation of radiation-induced toxicity.
Tablee 1, Patient, tumor and treatment characteristics.
Malee / female Agee (year) Tumorr stage
GTV V
Tumorr location
Electivee nodal field Involvedd field (60.8 Radiotherapyy dose Chemotherapyy prioi
median n range e
stagee f (IA/ IB) stagee ll(IIA/IIB) stagee III (I1IA / 1MB)
pre-radiotherapyy (cm3) mean range e central l
peripheral l
upperr lobe / middle lobe / lower lobe ++ boost to a total dose of 70 Gy -- 94.5 Gy)
too GTV (Gy) mean range e
rr
to radiotherapy / radiotherapy alone
63/19 9 74 4 48-88 8 266 (12 /14) 244 (2 /12) 4 2 ( 2 6 / 1 6 ) ) 113 3 2 - 9 0 1 1 48 8 34 4 6 3 / 7 / 1 2 2 20 0 62 2 74.8 8 60.88 - 94.5 4 / 7 8 8
Itt has been hypothesized mat regression of central tumors is likely to result in reperfusion (Markss 2000b). Consequently, we classified tumors in 2 groups according to their location (Tablee 1). A central lung tumor was defined as a tumor involving the hiiurn and/or mediastinum byy either the primary tumor or metastatic lymph nodes (Marks 2000b).
Forr all patiënte 3D-conformaJ RT plans were created including either an elective nodal field irradiationn or an involved field radiotherapy (Table 1), encompassing the primary tumor with lymphh nodes pathologic at mediastinoscopy, on CT scan according to the 1 cm diameter criterionn or, when PET was available, all lymph nodes showing 18FDG uptake. Only four patientss received chemotherapy, which was administered at least 6 weeks prior to radiotherapy. .
Standardd RT regimen consisted of 70 Gy delivered in 35 fractions and 7 weeks in 36 patients. Forty-sixx patiënte were treated within the context of an ongoing phase l/l I dose escalation trial
ChapterChapter 4
(Belderboss 2001; dose range 60.8-94.5 Gy / 2.25 Gy per fraction / overall treatment time 6 weeks).. Patients were entered in this trial after an informed consent was obtained and the trial wass approved by the hospitai's ethics committee.
Pulmonaryy function tests
PFTss were performed using Jaeger Masteriab equipment (Wurzburg, Germany). Data were collectedd at baseline (within 2 weeks prior to RT) and at 3-4 months follow-up. For this study thee forced expiratory volume in 1 second (FEV,) and the transfer factor for carbon monoxide 0"Lrco)) were analysed as these PFTs are most reported. FEV-, was measured with spirometry. TL,coo was determined using the single breath method. The transfer factor was corrected for thee actual hemoglobin level (Hb) in the peripheral blood (TLCoc) according to the formula TLpcoc ==
TL,co*(612 + Hb)/(1.7*Hb). PFTs were expressed as percentage of the predicted normal valuee according to Quanjer et al. (1983). Changes in PFTs were expressed as relative reductionss defined by the difference between pre- and post-RT value relative to the pre-RT value. .
3D-dosee computation
Aill patients had CT-based 3D-dose computations performed in our treatment planning system (U-MPIan,, University of Michigan (Fraass 1987)) as described previously (Boersma 1994). Tissuee innomogeneity corrections were based on an equivalent path length algorithm. To correctt for the effect of dose per fraction the local dose was converted to the normalized total dosee (NTD, Lebesque 1991) which is defined as the total biologically equivalent dose deliveredd in 2 Gy per fraction, using the linear quadratic model with an a/B ratio of 3 Gy (Van Dykk 1989). All doses reported hereafter are NTD-corrected unless stated otherwise.
Quantificationn of tumor regression
Alll CT scans (5 mm slice thickness) in RT treatment position were acquired during free breathing.. The gross tumor volume (GTV) was delineated on the pre- and post-RT CT scans usingg the appropriate level and window settings (lung and mediastinum). The GTV was definedd as the primary tumor solely except for hilar tumors in which the primary tumor is often contiguouss with adjacent lymph nodes. The relative tumor reduction was expressed as the differencee of the pre- and post-RT tumor volume divided by the pre-RT tumor volume.
Quantificationn of perfusion
SPECTT scans were obtained following injection of 4 mCi of 99mTechnetium-labeled macroaggregatedd albumin. All SPECT scans were made within one week of the companion CTT scan. Image acquisition and reconstruction were performed as reported previously (Seppenwooldee 2000). The interpretation of the SPECT scans is based on the concept of parallel-organizedd perfused subunits in the lung. The distribution of perfused subunits can be measuredd by the perfusion scans, in which the number of SPECT counts in a voxel is proportionall to the number of perfused subunits in that voxel. To compare the pre- and post-RTT scans quantatively, SPECT counts were normalized to the well-perfused low-dose (WPLD) regions.. This approach is based on the assumption that the number of perfused subunits remainss constant in the low-dose (no radiation effect) and well-perfused (no reperfusion effect) regionss (Kwa 1998a, Seppenwoolde 2000). Low-dose regions were defined as lung regions receivingg a dose less than 8 Gy, well-perfused regions as regions that contain voxels with a perfusionn of more than 60% of the maximum perfusion (before treatment, Seppenwoolde 2000). .
Thiss normalization allows quantification of perfusion changes. Matching of SPECT and CT scanss allows correlation of these perfusion changes with the locally delivered dose. In an 50 0
PulmonaryPulmonary function following high dose radiotherapy
earlierr paper by our group, the obtained dose-effect data for perfusion changes in individual patientss were averaged over the patient population and fitted with a sigmoid-shaped function accordingg to a logistic model with a Dso of 63 Gy and k of 1.7. The data were also fitted with aa linear relation with a slope S of 0.67% per gray (Figure 1).
Dose(Gy) )
FigureFigure 1. Dose-effect data for perfusion changes in breast cancer, lymphoma and NSCLC patients treated at the NetherlandsNetherlands Cancer Institute (triangles) and at Duke University (circles). Logistic and linear fits used for the analysis.
Variouss perfusion-related parameters were computed:
Measuredd lung perfusion
Thee perfusion homogeneity (PH) (Appendix II) was defined as the average number of SPECT countss over the whole lung and was calculated at baseline ( P H ^ ) and at 3-4 months post-RT (PHmeas-post).. The PH is 100% in case of homogeneous perfusion.
Thee measured perfusion reduction (MPR) was defined as the difference between the PHP** andd PHmeas-post relative to the PHf*8.
Predictedd lung perfusion post-radiotherapy
Forr each individual patient a post-RT perfusion scan can be predicted by combining the dose-effectt relation for perfusion changes (Figure 1) with the patient's pre-RT SPECT and individual 3D-dosee distribution. From this predicted post-RT scan, the predicted perfusion homogeneity (PHpred-post)) was calculated.
Inn analogy to the MPR, the predicted perfusion reduction (PPR) was defined as the difference betweenn PHf*6 and PHPred^K>sl, relative to the PHPre. It can be shown that in this case the perfusionn is effectively normalized to the average perfusion in the whole lung. If we approximatee the local dose-effect relation with a linear fit with a slope of 0.67% per gray (Figuree 1), the PPR is proportional to the mean perfusion-weighted lung dose (MpLD), which wass defined as the average perfusion-weighted dose to the lungs. For homogenous lung perfusionn the MpLD is (per definition) equal to the mean lung dose (MLD).
Thee relative reperfusion (rREP) was calculated from the difference between PHmeas st and
PHpred-postjj a g a j n relative to PHP re
ChapterChapter 4
Statisticall analysis
Forr the univariate analyses linear regression was used. Multiple factors were explored in a multivariatee analysis to investigate their association with reperfusion using logistic regression withh a stepwise backward elimination approach. At each step the least significant variable was leftt out when the significance level was above 0.05. The same procedure was performed to assesss the impact of various parameters on the reduction of PFTs (Table 3). P-values were not correctedd for multiple comparisons. Analysis was carried out using SPSS 9 (Superior Performingg Software Systems).
Results s
Pulmonaryy function pre- and post-radiotherapy
Thee average baseline values for FEV^ and TL.COC*©™ 60% and 69%, respectively (Table 2). Att 3-4 months post-RT FEV1 decreased on average by 6% while reductions of TL C O c were
largerr and on average 14%. In addition to reductions, improvements of PFTs were also observedd (negative values in Table 2). Thirty-eight percent of the patients experienced an improvementt of FEVy TL,coc improved in 2 1 % of the patients. Sixty-two percent of patients
withh an improvement of TLCoc experienced an improvement of their FEV,, while only 32% of
improvementss of FEV1 were associated with a simultaneous improvement of TL C O c.
Tablee 2. Pulmonary function tests. Lungg fraction parameter FEV-, , TLCÖC C mean(%) ) 1SDD (%) rangee (%) N N meann (%) 1SDD (%) rangee (%) N N BasalJfw w 60 0 20 0 288 to 121 82 2 69 9 23 3 200 to 128 74 4
Relativee reduction at 3 months 6 6 16 6 -344 to 41 82 2 14 4 19 9 -333 to 44 63 3
Averagee baseline (pre-treatment) pulmonary function parameters (% of predicted value) and their relativee reductions at 3-4 months following radiotherapy. The range and SD are also tabulated. Negativee values indicate an improvement of pulmonary function. N represents the number of patientss in whom PFTs were performed.
Relationn between lung dose and perfusion-weighted lung dose parameters
Thee MUD was on average 14.9 Gy (1SD 4.8 Gy) and was larger in the subset of central tumors ass compared to peripheral tumors (17.3 Gy versus 11.4 Gy, respectively) (right shift of datapoints,, Figure 2A). The average perfusion-weighted mean lung dose was overall 13.9 Gy (11 SD 4.7 Gy). For peripheral tumors the average MpLD was 11.6 Gy and similar to the average MUD.. In the subgroup of central tumors, the average MpLD was 15.7 Gy and lower than the averagee MLD (Figure 2A), due to the more inhomogenous perfusion patterns in these patients.
Measuredd and predicted perfusion changes
Thee perfusion homogeneity parameter is a measure for the average perfusion throughout the lungs.. The PHP"* averaged over all patients was 51.4% (1SD 9.2%). The PHme8S^)OSi was almostt identical and on average 51.6% (1SD 9.2%).
PulmonaryPulmonary function folfowing high dose radiotherapy
Thee measured perfusion reduction (MPR) was on average -1.2% (1SD 13.8%) indicating a smalll overall increase of perfusion, possibly due to reperfusion.
— — o o a> > Q Q c c 3 3 -o o 2 2 1 1 c c o o
1 1
s s 2 0 0 10 0 s s ,, , 0 0 *' *' ,, Peripherall , ' Centrall ' < ' ^ ' - . . : <p-<p-00 . y « 0 . 9 8 ' x oo QO r = 0.76 '' p < 0.001 0 0A.. Mean Lung Dose (Gy) B.. Mean perfusion weighted Lung Dose (Gy)
FigureFigure 2. A. Correlation between the mean perfusion-weighted king dose and the mean lung dose for central and
peripheralperipheral lung tumors. The dotted line is the line of identity. B. Correlation between the predicted perfusion reduction andand the mean perfusion-weighted lung dose. The solid line represents the regression line.
Basedd on the sigmoid dose-effect relation (Seppenwoolde 2000) (Figure 1) for local perfusion changess and a pre-RT SPECT lung perfusion scan, the PH post-RT was predicted. The mean PHPred-postt w a s 46.4% (1SD 8.1%) and thus on average 5.2% lower than the (actual) PHmeas"
fxx*.fxx*. The difference between the PHPre and the PHP"**-»^ relative to the PHPre is the predicted perfusionn reduction (PPR). This PPR was on average 9.5% (1SD 3.5%).
Thee correlation between the PHmeas st and the PHPred st was examined for each individual patientt (Figure 3A). For most NSCLC patients the (actual) PHmeas st was better than the PHPred-postt Approximately 80% of the individual data points are located above the line of identity.. This difference was ascribed to reperfusion and was independent of tumor location (centrall versus peripheral) (Figure 3A).
II 70 > 60 (|| 50 -5 -5 || 40 -O -O %% 30 -o -o »» 20 -c -c ££ io . Peripheral l Central l A.. Perfusion H o m o g e n e i t y " * * - " (%) ££ 40
II
30t» »
t?? 10 èè o 2 2 °> °> KK -10 -20 0 -30 0 --" --" --0 --0 Peripheral l Central l ' ' o o o o 1 1 0 0 0 0 OO # 0v
?s£> >
.. ' o * * " * .0 rr = PP = -00 32 0.003 3 30 0 5 0 0 6 0 0 70 0 B.. Perfusion Pre-RT (%)FigureFigure 3. A. Relation between the perfusion homogeneity1»**»1'' and the perfusion homogeneitf"***1 for centrally andand peripherally located lung tumors. The dotted line is the line of identity. B. Relative reperfusion as a function of perfusionperfusion homogene^ for the subset of central and peripheral King tumors. The regression line (dotted) represents thethe fit for the whole group.
ChapterChapter 4
Reperfusion n
Thee relative reperfusion was on average 10.9% (1 SD 13%). Several variables (PHPre, tumor volumeP"5,, delivered physical tumor dose, relative and absolute tumor reduction) were exploredd to evaluate their association with the relative reperfusion. In the univariate analysis, thee relative reperfusion was weakly (r=0.32) but significantly (p=0.003) correlated with PHPre (Figuree 3B). No other correlations were observed.
Inn a subgroup analysis according to tumor location (central versus peripheral) the correlation betweenn the relative reperfusion and the PHP"* was significant only in peripheral tumors (p=0.01)) (Figure 3B). In central tumors, there was a trend for a correlation (r=0.3, p=0.08) betweenn reperfusion and tumor regression (Figure 4).
60 0 400 -•• ~ . • » : **-.-;; * * vv • •• •• • •• •• • • • • • r=0.3 3 pp = 0.08 100 20 30 40 50 60 70 80 90 100 110
Relativee Tumor Reduction (%)
FigureFigure 4. Relative reperfusion as a function of the relative tumor reduction in 48 patients with centrally located tumors.
TheThe dotted line indicates the regression line and the solid lines represent the 95% confidence interval for the patient population. population.
Alll factors explored were included in a multivariate model. In the multivariate analysis on the wholee group, the relative reperfusion was, as in the univariate analysis, only significantly associatedd with the PHPre. Given the trend for an association between reperfusion and tumor regressionn in central tumors in the univariate analysis, the multivariate analysis was also performedd incorporating tumor location as a dummy variable (1-central, 0=peripheral). In this analysis,, only the PHPre was significantly associated with reperfusion in both central and peripherall tumor locations.
Estimationn of changes in pulmonary function tests
Differentt variables (Table 3) were first tested in a univariate analysis for their association with thee relative reduction of the PFTs. The relative reductions of all PFTs were strongly associated withh the PPR and MpLD. This is not surprising since the PPR and MpLD parameters are stronglyy correlated (Figure 2B). It can mathematically be shown that the PPR is proportional too the MpLD if the PPR is calculated using a linear dose-effect relation with slope S (see also Methodss section, Figure 1). Since we used a sigmoid-shaped dose-effect relation to calculate thee PPR, the coefficient of correlation between the PPR and the MpLD slightly differs from 1 (Figuree 2B).
Thee tumor volume, tumor dose, MLD, MPR and reperfusion were not associated with functionall outcome as measured by the PFTs (Table 3).
PulmonaryPulmonary function following high dose radiotherapy
Tablee 3. Variables assessed for an association with changes in pulmonary function tests
Tumorr volumepre Totall tumor dose Meann lung dose
Meann perfusion-weighted lung dose Predictedd perfusion reduction Relativee tumor reduction Measuredd perfusion reduction Relativee reperfusion rrFEV, , 0.60 0 0.56 6 0.24 4 0.033 0.01* 0.022 0.01* 0.077 0.02* 0.22 2 0.53 3 öTy?o« « 0.41 1 0.07 7 0.05 5 0.01* * 0.02* * 0.022 0.05 0.92 2 0.70 0
Correlationn of different variables with the relative reductions (rr) of F E V , , TLiCOc- The p-values are
tabulated.. Statistically significant p-values in the univariate analysis are in bold and significance in thee multivariate analysis is indicated in bold and with asterisk.
Figuree 5 summarizes in different panels the correlations between the relative reductions of PFTss and the MLD (panels B) and the MpLD (panels C). For comparison, the correlations (r=0.588 to 0.69) between the relative reductions of TL C O c and FENA, and the MLD were also
displayedd for a group of reference patients (breast cancer and lymphoma) who received incidentall partial irradiation of their (healthy) lungs (panels A, adapted from Theuws et al. (1998)).. The slopes of the regression lines indicate a 1% reduction of TLCoc and FEV^ per gray
(Panelss A).
Breadd Cancer, Lymphoma Lung Cancer Lung Cancer
m» »
r-0.24 4 pp • 0.03
00 10 JO JO 0 10 SO JO 0 1 0 Ï Ö J O
A.. Mean Lung Dos* (Sy) B. Mean Lung Dose (Gy) c. Mean perfusion-weighted Lungg Dose (Gy)
Figuree 5. The relative reductions (rr) of TLCOc and FEV1 as a functhn of the mean lung dose for reference patients with
breastbreast cancer and lymphoma (Panels A, adapted from Theuws et al. (1999)) and studied NSCLC patients (Panels B). ForFor the NSCLC patients, the reductions are also displayed as a function of the mean perfusion-weighted lung dose (Panels(Panels C). Data are binned for the display. The error bars represent standard error of the mean. The regression lineslines are dotted.
e»H H I15> > «•> > y y '' <«> rr . 0.69 =-00 000: ' ' m m
F F
o» »-+--i » ) )
A A
HO OA A
r«0.M M P"" 014ChapterChapter 4
Inn the studied NSCLC patients, only the MpLD and PPR but not the MLD were weakly (r=0.24 too 0.31) correlated with the reductions of TLCOc and FEV, (Figure 5, Table 3).
Inn a subsequent multivariate analysis both PFTs remained significantly associated with the MpLDD or PPR. Furthermore, the relative reduction of FEV1 was inversely correlated with the relativee tumor reduction (p=0.02). For TLlcoc the association was borderline (p=0.05).
Discussion n
Changess of PFTs
Manyy investigators have quantified PFTs following thoracic radiotherapy as their decline has beenn con-elated with clinical symptoms (Choi 1985, Rubenstein 1988, Marks 1997, Theuws 1998,1999,2000,, Fan 2001b). In the current study, the lung volume parameter FEV, declined onn average by 6% (Table 2). These figure compares favorably with the results of other studies inn lung cancer in which decreases in the range of 11-20% were found (Choi 1985, Rubenstein 1988,, Fan 2001b). We measured an average decrease of TLiCoc of 14% which is again at the lowerr limit within the range of reported values (approximately 17-28%) but twice as large as thee reduction of the lung volume parameter.
PFTss and MLD
Soo far, the ability to predict the magnitude and direction of changes in overall pulmonary functionn as measured by PFTs has yielded rather disappointing results in patients with NSCLC (Fann 2001 a,2001 b). This is in contrast with the prediction of PFTs in patients with healthy lungs undergoingg incidental partial lung irradiation, in a group of 81 breast cancer and lymphoma patientss Theuws et at. (1998) showed mat the mean lung dose (MLD) (a combination of radiationn dose and irradiated volume) is the strongest predictor for changes in PFTs. They foundd a relative decrease in PFTs of approximately 1% per gray. In NSCLC patients a dose-volumee parameter like the MLD does not relate to the reduction of PFTs (Figure 5, panels B andd Table 3). In fact this is not surprising as the MLD implicitly assumes that each part of the lungg contributes equally to the overall lung function (Theuws 1999). For patients with NSCLC thiss assumption does not hold due to the presence Of cancer but also due to pre-existing lung diseasess associated with unequal perfusion throughout the lungs.
PFTss and perfusion
Basedd on the experience in thoracic surgery where the percentage of perfusion of resected lungg correlated with the percentage reduction of PFTs, (Bolliger 1995, Giordano 1997) it can bee postulated that the sum of regional perfusion changes post-RT might correlate with the changee in overall pulmonary function as measured by PFTs (Fan 2001a, 2001b). Following thiss assumption, we have studied various perfusion parameters in relation to PFT changes. Ourr group (Seppenwoolde 2000) and the group from Duke University (Garipagaoglu 1999) havee combined full 3D SPECT scans and dose distributions and established comparable dose-responsee curves representing the radiation-induced dose-dependent reduction in regionall lung perfusion (Figure 1). Based on this dose-effect relation and each patient's individuall dose distribution and pre-treatment SPECT scan a post-RT perfusion homogeneity cann be predicted. In most NSCLC patients the measured perfusion reduction was smaller than predictedd based on the dose-effect relation and this was attributed to reperfusion.
PulmonaryHmctionPulmonaryHmction following high dose radiotherapy
Reperfusion n
Wee have previously reported on the occurrence of reperfuston in NSCLC (Seppenwoolde 2000).. In the current study we further examined factors associated with reperfusion. Reperfusionn was found to be correlated with the perfusion pre-RT (Figure 3B) suggesting that lungg regions that are relatively less perfused are more likely to shew reperfusion following RT ass compared to better perfused regions. Only in central tumors, reperfusion shews a trend for ann association with tumor reduction. This is in agreement with the hypothesis that perfusion recoveryy is more likely to occur in patients with central tumors. These tumors often cause obstructionn and encasement of pulmonary vessels thereby reducing tile regional perfusion (Markss 2000b).
Parameterss to estimate overall changes In PFTs
Choii et at. (1994) reported that 60% of patients with central tumors experienced an improvementt of their PFTs. Marks et at. (2000b) observed in 45% of patients with a central tumorr and adjacent hypoperfusion an improvement of PFTs while an improvement of PFTs wass demonstrated in only 10% of the other patiënte. Given these observations we hypothesizedd that tumor regression (which was not quantified in the referenced studies) might affectt overall pulmonary function indirectly (through reperfusion) or directly (de-obstruction of airways,, reinftatton of collapsed lobes).
Inn our analysis the predictive value of various tumor, dose and perfusion related parameters forr changes in PFTs was tested (Table 3). In the uni- and multivariate analyses, the perfusion-weightedd equivalent of the mean lung dose, the mean perfusion-weighted lung dose was significantlyy associated with the changes of PFTs (Table 3, Figure 5 panels G). Although the correlationn coefficients were small, the correlation and significance were consistent for all PFTs.. Similar weak but significant correlations were observed between the predicted perfusion reductionn and reductions in all PFTs. It is unclear why in the studies by Marks the perfusion-weightedd parameters did not better predict PFTs as compared to the dose parameters (Fan 2001a,2001b). .
Thee measured perfusion reduction was not correlated with the reductions of PFTs. As the measuredd perfusion reduction differs from the predicted perfusion reduction by the reperfuston,, this suggests that reperfusion does not contribute to an improvement of pulmonaryy function. This is confirmed by the absence of association between reperfusion and
anan improvement of PFTs (Table 3). In particular, no correlation was observed between
reperfusionn and an improvement of TLCOc. We hypothesize that radiation induces damage to thee alveolar/capillary membrane so that a (potential) perfusion recovery does not necessarily
franslatefranslate into an improvement of overall pulmonary function. It might also be hypothesized that thee duration of a perfusion deficit might be critical for the probability of recovery. To our
knowledge,, no data are available on this issue.
Althoughh reperfuston did not translate into an improvement of overall PFTs, tumor regression didd correlate significantly with an improvement of both FEV, and TL.coc- This observation suggestss rather a direct impact of tumor regression on PFTs than an indirect impact through reperfusion. .
Wee found that changes in PFTs post-RT can be best estimated by perfusion-weighted dose parameters.. However, it should be noted that the correlation coefficients are small. Although perfusion-weightedd parameters provide a better estimate of functional outcome after high dose radiotherapyy of NSCLC than pure dose parameters, it remains difficult to accurately predict PFTss for an individual patient It is possible that the incorporation of biological factors like TGF-p** In addition to patient-specific factors (age, gender, smoking habits) in a multi-parameter
ChapterChapter 4
modell might further improve our ability to predict PFTs. Inclusion of biological factors (Fu 2001) inn conjunction to dose parameters provides a better prediction of radiation pneumonitis. Inn summary, tumor regression following high-dose radiotherapy of NSCLC can lead to an improvementt of pulmonary function. Reperfuston does not translate into an improvement of pulmonaryy function tests. The perfusion-weighted dose parameters revealed the best availablee parameters to estimate pulmonary function post-radiotherapy.
